RFP answering

Open Admin, in the left pane, then its Advanced features tab: the RFP card there shows whether RFP is on for your company. Where the card offers it, a Zime company admin turns RFP on for everyone in your company and confirms with Turn on, and RFP then appears as a tab of Workflows, in the left pane. The left pane shows Admin to a Zime company admin only, and only a company admin can open its Advanced features tab, so anyone else asks one of them.

Turning RFP off works the same way, with Turn off: RFP leaves Workflows, which leaves the left pane too while it holds no other feature, and RFP's pages say that RFP is turned off, with Open Advanced features for a company admin and whom to ask for anyone else, and the Get started tab above that message still opens this guide. It deletes nothing: your projects, the Library and your credit history are kept, and turning RFP back on brings them back. A project already in Trash is still removed for good after its 30 days there, and a Generate run already going still finishes.

Open Projects and choose New Project. Give the project a name, attach the questionnaire your customer sent, and fill in what you know about it: the project type, the due date, a description, the client, and the co owners who should manage it with you. The same dialog sets the Knowledge scope and whether Approved answers enter the Library. Then choose Upload and map.

Everything but the file can be changed later. While the project is under Active or Completed, a project owner or an RFP Admin changes its details and the Library setting from About, on the row menu of the projects table, and, once the import is confirmed, its Knowledge scope with the Knowledge button beside the project name on the Questions tab. About also holds Instructions for answering: text given with every question of the project when its answer is generated. While it holds text, an answer approved in the Library is not reused as it stands: it is rewritten to follow those instructions, and the rewrite is told to keep its facts, figures, names and claims. When the approved answer does not hold what they ask for, the question is answered as if the Library had no answer for it.

The projects table lists what you are on, under Active, Completed, Archived and Trash, with a search and sortable columns. An RFP Admin sees every project of the company; a Manager sees the projects they own or have been assigned work on.

A project starts from one file: the questionnaire your customer sent you. Three formats are read: an Excel workbook (.xlsx, the macro enabled .xlsm included), a Word document (.docx) and a PDF (.pdf). Upload it exactly as you received it, with nothing stripped out.

A PDF can be uploaded as it is. To have the answers written into the document itself, convert it to Microsoft Word first, by opening it with Word or with the Adobe PDF to Word converter, and upload the .docx.

Whatever the format, nothing is imported until you have reviewed what Zime detected and confirmed it.

Each file you upload to start a project uses 2 credits, charged up front to the person who uploads it, first to their own monthly AMA credits and then, if your plan has a company pool, to that pool, as an answer is: Zime checks those credits before it reads the file, and an upload they cannot cover is refused, with the reason on screen. An upload that fails, such as a file that cannot be read, costs nothing.

Nothing is imported until you confirm it. Zime reads the workbook and works out which sheets hold questions, which column holds the question text, which column or columns hold a compliance dropdown, and which cell or cells each answer belongs in. That map is shown back to you on the review screen, over a preview of your own sheet. A sheet, a row or a column your customer's workbook hides is not shown there and cannot be marked, and nothing on it is imported, so nothing on it is answered.

A Word or PDF questionnaire gets the same confirm first review, and it opens on the page view: your document, drawn as the customer sent it, with a badge beside every row that carries a mark, Zime's or yours, such as S for a section, Q for a question and I for a row left out. What the letters mean, in the Mark options pane, explains every letter. Click a paragraph to mark it. Drag across paragraphs, or click the first one and Shift click the last, to select a range: one mark then covers every row in it.

To mark every row that looks like one you marked, select that row, open the three dots menu on the toolbar and choose Auto Identify: it lists the rows that share its formatting, leaving out the ones that already show that mark and the ones you marked otherwise. Untick the ones you do not want, and Apply marks the rest.

On that screen you can change what any column is, leave a sheet out with Ignore this sheet in the menu on its tab, bring in a sheet Zime left out with This is a questionnaire in the same menu, and edit a question, its answer type and any character or word limit in Question details. On a workbook you can also mark a column Context, so each of its cells is added to the question on its row, and mark a row INSTRUCTION, so its text is given with every question of that sheet when its answer is generated, or with every question of the workbook when the sheet has no question to answer, as a cover or instructions sheet has. Zime proposes the text of such a sheet as instructions, under This sheet in the Mark options pane, and none of its text is given as instructions until you take the proposal: Use as instructions marks its rows INSTRUCTION, bringing a sheet Zime left out into the import first, and Not instructions leaves the sheet as it is. Where the question column of a workbook is merged down beside drill down options listed under a question, each option with an answer cell of its own comes in as a sub question of that question; elsewhere, mark such options Level 1 Sub Question yourself. Under This sheet, in the Mark options pane, Zime lists what wants your decision, such as the rows of a Word document it was unsure about, the Word headings that hold no question (which are not imported as sections) and the cells the customer filled in, so the rows worth a second look come to you rather than being buried.

Import stays blocked while anything about the map would make the import wrong, for example two columns marked Question, a sheet with no column marked Question (a sheet whose every row of text is marked INSTRUCTION or Ignore needs none), one cell set to take both an answer and a compliance choice, or no questionnaire sheet to import from. The line under Import names what it is waiting on each time, and the banner over the preview says what to do about it: over a sheet with no column marked Question it offers the columns that could hold the questions, and Bring back Zime's suggestions where Zime's own reading of that sheet has a Question column. A file that holds nothing to answer at all is refused there too, rather than imported as an empty project. A row that would come in as a question with no words does not block the import: a warning names it. Once the import goes through, the project opens on its Sections tab.

Not in your customer's workbook or Word document. Either comes back as their own file, and the only content Zime writes into it is answers and compliance choices: into the answer cells of a workbook, and in a Word document under each question, into its answer cells, or in place of an answer you marked ANSWER. The question cells and question paragraphs stay exactly as they are, unless you point a question's answer at its own cell.

If a project owner or an RFP Admin edits a question, with Edit question on its card or in Question details on the review screen, Zime answers that wording, the project shows it, and the Library entry an approval makes carries it too. Your customer's workbook or Word document keeps their own wording, exactly as they wrote it.

A PDF comes back as a Word document. With Questions and answers only off, it is your customer's whole PDF, each question in their words; the answer of a question you added goes at the end, under "Additional questions and answers", with the question as you wrote it. With it on, a table that came in as a table of questions, with its own table handle, reads as your customer printed it: a question you edited reads there in your customer's words, and one you added reads as you wrote it. Anywhere else in that document each question reads as it reads in the project, your edits included.

Before the import is confirmed: select the row on the review screen and choose Ignore in the Mark options pane or in the three dots menu on the toolbar. Nothing is imported from that row, so it is never answered and never counted. Zime already leaves out what it can tell is not a question: in a Word or PDF document such a row, a table of contents line for example, carries the I badge, while on a workbook a row Zime leaves out carries no badge at all. So check a row's badge before adding your own Ignore. To leave out a whole sheet, choose Ignore this sheet in the menu on its tab. Clear, under More in the Mark options pane or in the three dots menu on the toolbar, takes your mark off a row you marked.

After the import, a project owner or an RFP Admin opens Configuration, marks the row Ignore, then chooses Apply changes and Apply: the question leaves the project and goes, with its answers, to Removed by Apply on the Sections tab, where Restore brings it back. Delete, in the three dots menu on a question card, removes the question and its answers for good, and its sub questions move up to its parent; a question that came from a row of the file comes back as a new question at the next Apply changes unless that row is marked Ignore.

None of this edits your customer's file: a question you left out simply gets no answer written for it. A PDF comes back as a Word document. With Questions and answers only off, it is your customer's whole PDF, a question you left out as they printed it, with no answer. With it on, a table that came in as a table of questions, with its own table handle, has a row you left out as your customer printed it, with no answer, and anywhere else a question you left out is not in that document.

Zime leaves out, with the I badge, the paragraphs it reads as asking for nothing: a table of contents, a paragraph with no letter or digit in it, such as a line of dots or underscores, and a sentence that ends in a colon and asks for nothing, together with the bulleted or lettered list under it. An item of that list that says shall, must, should or another word of obligation still comes in as a question.

Under This sheet, in the Mark options pane, Zime lists the rows it left out with each such list as rows it was unsure about, and Show points at them in the grid. To answer a row Zime left out, select it and mark it QUESTION.

Zime reads each table drawn in a PDF as a table of its own. A table with a header row, 3 rows or more under it, a column of sentences, and a column left empty or nearly empty to answer in comes in on its own as a table of questions: a row with its sentence is a question, with the ID its row gives it where the table has a column of IDs, and a row that only titles the rows under it is a section. Its answers go into the columns Zime proposes for them: usually each column with a header right of the sentences that is empty or nearly empty, together with the named column just right of the first of them, filled or not. Table Properties shows them, and you can change them there. Every other table stays in the page, its lines read as the rest of the page is. Each table has a table handle on the page view, and the handle of a table that stayed in the page says it is not imported.

On the page view, every line of the PDF is in a row you can mark, as a section, a question, Ignore or any other mark the row takes, a paragraph or a list item being one row. What Zime reads as repeated from page to page, such as a running header, a page number or a table header row printed again on a new page, is not a row. A row Zime read as neither a question nor a section carries the I badge, so a question Zime did not see is still on the page to mark QUESTION.

Zime cannot read words that are a picture, as the pages of a scanned PDF are: put such a file through OCR first, or ask for the version of the document that carries its own text, and import that.

A PDF project that was already in Zime when Zime began to read PDFs this way keeps reading as it did, with no table handles on its page view: its review and Table Properties stay as they were, Attach original file reads its file the same way, and its export is the Word document every PDF project exports.

Select the table on the page view, by its table handle or one of its cells, and open Table Properties, with TABLE PROPERTIES on the toolbar or Table Properties in the Mark options pane. As on a table of a Word document, it sets the header row, the Question column and each Answer column with its answer type, Text, or Dropdown or Scale with its values, and Save applies it to the table.

A table that stayed in the page comes in the same way: Save in Table Properties brings it in as a table of questions, and while it is in, its lines are read only as its rows, any mark you made on them on the page kept for when it goes back. If Zime cannot read it as a table of its own, the dialog says so and the table stays in the page, for its rows to be marked there. To take a table setup off, select its table handle and choose Clear, under More in the Mark options pane or in the three dots menu on the toolbar: a table you brought in goes back to the page, and any other table is read again as Zime first read it.

In a PDF with two tables or more, set up one table first, then choose Auto Identify Tables, above the page or in the menu beside TABLE PROPERTIES on the toolbar: every other table that looks the same, with the same columns under the same header words, takes that setup, a table that stayed in the page coming in with it when Zime can read it as a table of its own. So do the parts of the table you set up that have no header row of their own, such as the groups of a table whose header is printed on its first page only. A table already set up is left as it is, and a message names the tables that took the setup.

Choose Clear Sheet on the bar above the sheet. It asks first, then removes the marks on the sheet you are looking at; the other sheets stay as they are.

The dialog offers two choices. Where it can tell what each would do to your sheet, it says so, and a choice that would change nothing is off, with the reason under it. Clear to a blank sheet removes every mark on the sheet, whoever made it, Zime's suggestions and the column marks included: nothing on that sheet is imported until you mark it. It keeps the header row, and the answer type and Applicable to Rows set for each column in Answer Configuration, which apply again once you mark that column. On a workbook, start with the columns: QUESTION on a cell that holds a question's words makes its column the Question column and marks that row, and ANSWER on a cell where an answer goes makes its column an Answer goes here column. Then choose Auto Identify for the rows like the one you marked.

Bring back Zime's suggestions removes every row mark on the sheet, whoever made it, and shows Zime's suggested sections and questions again; when Zime has none to show, it says so. When the line under this choice names them, it also removes the header row and the column roles and answer configuration set on that sheet, and Zime reads the sheet again. Either choice also takes off what was set on the sheet's rows, in Question details for example, and opens the decisions under This sheet again; Rules by value stay as they are. A sheet cleared to blank says so in a line above it, and Bring back Zime's suggestions on that line opens Clear Sheet on that choice.

Undo takes a clear back. Zime keeps what the last clear of a sheet removed: the marks, whoever made them, the decisions under This sheet, the header row and the column roles and answer configuration it took, and whether the sheet was blank. Undo, the curved arrow on the bar, puts all of it back while the page stays open. Undo clear does the same, on the line above a blank sheet and in the menu of the sheet's tab, and it is still there after you reload the page, and for a colleague. Hover over Undo clear to read what it puts back, the header row chosen for the sheet included. A row marked after the clear keeps its new mark. The exception is a row on or above a header row that Undo puts back: its mark goes, and the message after Undo says how many did. A header row, column role or answer configuration set after the clear stays as it is. Zime keeps the clear until it is undone or the sheet is cleared again.

Clear Sheet is off on a sheet Zime left out of the import that holds no marks, because there is nothing on it to clear. A line under the bar says so. To import from such a sheet, mark its header row, or choose This is a questionnaire in the menu on its tab and then mark its header row. In a Word or PDF document, choose Grid, then This is a questionnaire in the menu on the table's tab. A table of a PDF that stayed in the page, its table handle saying it is not imported, has no tab there: select its table handle and open Table Properties.

On a project already imported, clearing a sheet changes nothing until you choose Apply changes. Read its preview before you choose Apply: a row that is no longer a question leaves the project and goes, with its answers, to Removed by Apply on the Sections tab, where Restore brings it back.

It means the customer locked that sheet with sheet protection before sending you the file. The lock changes nothing about answering: every question on a protected sheet is detected and answered normally.

At export nothing is written into a cell that lock holds shut, so that answer does not appear in the workbook your customer gets back. It is not lost: it stays on the project, where you can still read it on the question. The import review screen points at the rows whose answer cells are locked, so you can see which ones they are before you import.

The workbook comes back to your customer with its protection exactly as it arrived. The check every export passes before it is handed over proves that the protection, like everything else the export did not change, is identical to the original. If you want an answer written into one of those cells, the customer has to unlock it in their file, and you start a new project from that file.

Yes. An answer column whose cells carry a dropdown list in your customer's file takes one of that list's values, spelled exactly as the customer spelled them, whether the column is marked Answer goes here or Compliance dropdown, unless its header names values of its own or you give the column another answer type yourself. Answer Configuration, which the gear on the bar above the sheet opens, shows it as a Dropdown with the file's values, and Auto Identify drop downs, in that dialog, also reads the values of a list the file keeps on another sheet.

Every question row takes a value from that list, a row the customer's list does not reach included, unless the sheet has a Requested Answer column that Zime reads, which then decides row by row, or you narrow the column with Applicable to Rows or Rules by value in Answer Configuration, leave that cell out of the answer, or give the row an answer type of its own. A header that names its own values keeps them: a score from 0 to 3 stays that scale, on the rows the customer's list covers, and a header asking for Yes or No offers those two answers, spelled as the header spells them.

When the value Zime generates for such a cell is not one of the list's values, the cell is left empty and the card names it as outside the list. Where the list explains its values in its own words, the review offers Use these labels under This sheet: the explanations become the values Zime shows you, and the export still writes the customer's own entries. Where the list behind an answer column is broken in the customer's file, This sheet says so, and Type the values opens Answer Configuration to add them.

The answer types Compliance dropdown (customer values) and Compliance dropdown + text answer are offered for a single question wherever one of the sheet's answer columns is a dropdown with values: in Question details on the review screen, and on a question you add after the import. A project owner or an RFP Admin changes the answer type of a question from the file with Edit Answer Type in the three dots menu on its card.

Yes, with Rules by value in Answer Configuration, which the gear on the bar above the sheet opens. In Type column, choose the column whose value tells the rows apart, such as the column where your customer wrote the kind of answer each row asks for; with no rule saved yet, it opens on the customer's Requested Answer column, where Zime reads one. Zime lists each value of that column with the number of question and information rows that hold it. Tick, for each value, the answer columns its rows fill, then choose Save. You set it once for the whole sheet, with no row numbers.

Each value is read in one of three ways. With columns ticked, its rows fill those answer columns. With Rule ticked and no column, its rows are information only: they come in marked Information only and are not answered. With no rule, its rows keep the answer cells they have now. Only the sheet's answer columns can be ticked, so a column you set to Not used is never filled by a rule. A dropdown or scale column with no values yet gives those rows no cell there until you add its values in Answer Type, and the section says so on the value and under the rules.

Where a rule sets a column, the Applicable to Rows field of that column in Answer Configuration says so, starting "Set by a rule" and naming the column and the values. The rules take effect when you save them: before the import on the review screen, and after it through Apply changes, whose preview shows what they change before anything is written. Rules by value are for workbooks: a Word or PDF questionnaire does not offer them.

They are imported as the answers of those cells. When you import the file, the text the customer already wrote in an answer cell becomes that cell's answer on the question, marked From the file on its card, and so does a value the customer chose in a dropdown cell. Nothing is generated or charged for it, and while those cells hold the customer's text, the export leaves them exactly as the customer wrote them.

A cell that holds only a placeholder, such as a dash, or a value that is not one of its dropdown list's values, is not imported. It stays an ordinary answer cell, and a generated answer replaces what is there. Where Zime reads pre filled text as a placeholder or a note rather than an answer, the review screen says so and offers Not answers (Not an answer, for one cell): those cells are then not imported, Zime answers them and the export writes over them. Until you choose that, or if you choose Keep them as answers (Keep it as the answer), their text is imported like any other.

Generate all answers and Generate all answers in this section never replace a cell that came from the file. On a row where some cells came from the file and others are empty, they generate the empty cells and keep the rest, and the card offers the same for one question with Generate empty cells. When Zime leaves such a cell empty on purpose, because your Instructions for answering tell it to (for example "leave the Comment empty when Fully Compliant"), that is no failure: Generate all answers and Generate empty cells stop offering that cell, and Edit or Refine fills it. Until those cells are filled the answer cannot be approved as the file gave it: Approve on its card names the empty cell, and the bulk approvals leave it and say how many they left. Once a person edits the answer, it can be approved as it stands, and Generate all answers no longer fills its empty cells. On an answer that holds cells from the file, Refine reads Replace with a generated answer, because it asks for every cell again, the imported ones included.

To keep the customer's pre filled text out of Zime altogether, choose Leave pre filled cells as the customer filled them, under This sheet in the Mark options pane, before you import: the sheet's pre filled text cells are then not imported and never written. The values the customer chose in dropdown cells stay imported. For a single cell, select it, open the three dots menu on the toolbar and choose Leave this cell as the customer filled it.

An answer column you add after the import, with Apply changes or Edit Answer Type, brings in its pre filled cells the same way, and the Apply preview counts them. A dropdown cell you move onto a cell the customer filled takes its value with it, and that value is written there over the customer's value.

A Word questionnaire's written answers come in when you mark their paragraphs ANSWER on the review screen, or choose Mark Rest as Answer, under ANSWER: each comes in whole as its question's answer, and its card says it is from the file. The export leaves those paragraphs as the customer wrote them unless you approve a changed answer, which then takes their place; once such an answer is marked Needs re read after its question changed, a Generate run writes it again whole. A table in a Word document is read cell by cell, as a workbook is, and so is a table of a PDF that came in as a table of questions, with its own table handle: an answer your customer wrote in one of its answer cells comes in as that cell's answer. No other answer in a PDF is imported.

Everyone in your company has one RFP role for the whole RFP workspace: Admin, Manager or None. An RFP Admin is whoever has been given the Admin role under RFP roles, on the Members tab in RFP, which only an RFP Admin is shown; everyone else is a Manager unless set to None, and only an RFP Admin changes a role there. An RFP Admin sees and can manage every project. A Manager sees the projects they own or are assigned work on. None closes the RFP workspace to that person.

Inside a project there are owners, authors and reviewers. Owners are set on the About form and run their project: they assign people, generate, export, archive and trash it. Being an owner does not override the reviewer: while a question has a reviewer, only its reviewers can approve it, and an owner or an RFP Admin who wants it approved changes or removes the reviewer first; sending an answer back is the reviewer's alone. Authors write answers. Reviewers approve them or send them back. Being made an author or a reviewer is what puts a person on a project: there is no separate member list, and removing their last assignment takes them off it again, unless they own it or are an RFP Admin.

Open the Sections tab. Each section row shows its Authors and its Reviewers: choose the plus beside either, or Manage Users in the three dots menu at the end of the row, to open Manage Users on that section. To set several sections in one save, tick them and choose Manage Users on the bar above the table. On the Questions tab, tick cards and choose Users on the bar to assign single questions, or choose Manage Users in a card's three dots menu.

Manage Users has an Authors tab and a Reviewers tab, each with four modes. Each tab opens listing the people set on that section or question itself, so one that takes its people from above opens empty, and with several ticked the list holds everyone set on any of them. Assign adds the people listed without removing anyone already on it, Reassign replaces who is on it with the people listed, and Remove takes the people listed off it, so leave in the list only the people you want removed; an empty list clears it. With several ticked, the mode applies the whole list to each of them. Assign and Reassign can give the people they add a due date and time, and Update due date gives the people already listed a new one, adding or removing nobody. Beside each person listed is the due date they hold, or "different due dates" when several are ticked and their dates differ; a person with no date of their own shows none and is due when the project is. When a section or question takes its people from above, those people are shown under the list, read only, each with the date they hold on the section above. Save writes only the tabs you changed.

A question takes its section's people unless it has its own, and a subsection takes its parent section's; the first people you set on one of them replace what it took from above, so add those too if they should stay. If you set people on a section while some of its questions carry their own, Zime asks before it saves anything: Reset puts those questions on the section's people and saves, and Cancel saves nothing. Only a project owner or an RFP Admin can assign.

Whoever is made a question's author owns its text. An author, a project owner or an RFP Admin can generate, refine, edit or write the answer to it. A reviewer can only do that while nobody has been made its author; the moment an author is assigned, the reviewer approves or sends back and does not rewrite.

Approval works the other way round. While a question has a reviewer, only its reviewers can approve it, and an owner or an RFP Admin who wants it approved changes or removes the reviewer first. A question with no reviewer is approved by an owner or an RFP Admin. Sending an answer back is the reviewer's alone.

On a card, a control you may not use is either disabled, and says why when you hover it, or not shown at all: Send back shows only to the question's reviewer, for example, and the card's three dots menu offers Manage Users, Edit question, Edit Answer Type, Move to section and Delete only to a project owner or an RFP Admin. Zime refuses the same request from anywhere else.

Choose Generate all answers on the Questions tab to draft every unanswered question of the project in one run, or Generate all answers in this section in the three dots menu of a section, on the Sections tab or in the list of sections on the left of the Questions tab, for one section and its subsections. To draft only some questions, tick their cards and choose Generate selected. Generate all answers and Generate all answers in this section skip what is already answered, so choosing either again only fills the gaps, such as a question with no answer or the empty cells of an answer imported from the customer's file. The exception is an answer marked Needs re read after its question changed: a run writes it again, as a new version that counts like any other, keeping every cell that came from the customer's file, and leaves alone one whose every cell came from the file. Both leave out the questions of a frozen section and the questions marked No answer found, and Generate all answers takes the latter too when you tick Also retry questions with no answer found beside it. Pause, on the running banner, stops a run once the answers already under way are saved, and Generate all answers then picks up the questions still waiting.

A single question has Generate answer on its card. Before generating you can type into the card: Save as my answer keeps your words as the answer, and Generate answer uses them as guidance for the generated one. On a question with two or more answer cells, or with a dropdown or score cell even when that is its only answer cell, the card shows a field for each cell for your own answer, a dropdown or score cell offering its own list, and guidance goes in its own box, Guidance for the AI (not an answer field). What you type there only steers Generate answer: it is not your answer, and it is never saved or exported as one.

An owner's or an RFP Admin's Generate all answers covers every question. Anyone else's run drafts only the questions they may write; the others are left for someone assigned to them, an owner or an RFP Admin, and the running banner says how many.

A run's answers are charged to the person who started it, and the run stops when their AMA credits run out, at their monthly credit limit or when your company's pool cannot cover the next answer. It says so, and the Credits topic explains what you then see.

Each answer cell of the row gets its own answer. The card shows one field per cell, in the order of the columns, each named by the column's header and its cell, for example Comment · G12. A text cell shows its text and a dropdown cell its value; once the question is answered, a dropdown cell's field also holds its own list, and choosing a value there saves it for that cell alone.

Before the first answer, the card shows the same fields to fill in: a box for each text cell and the cell's own list for each dropdown cell. Save as my answer keeps each in its own cell, and what you type in Guidance for the AI (not an answer field) is sent with Generate answer as guidance and is never saved or exported as the answer.

Generating asks for every cell it fills at once and writes one text per cell: one cell's text is never copied into another, and when the answer comes back with the same text for two cells, the first keeps it and the later one is left empty. A cell the answer leaves empty reads Empty on the card, followed by what fills it, such as Refine or Edit. A dropdown cell is also left empty when the value Zime chose is not one of its list's values, and the card names that cell as outside the list.

Edit opens one editor per text cell, and a save keeps every cell you did not change. An answer can also hold one text for the whole row: the export then writes that one text into each of its text cells, as its card shows.

Where a sheet repeats its answer columns, such as a dropdown and a comment for each of several parts, the review screen, the card and the export name the cells each row is written into, as your marks and Answer Configuration leave them.

Yes, for the questions you tick. Tick their cards and choose Generate selected. When some of them hold answers or cells imported from the customer's file, Zime asks first and says how many. Replace them (Replace it, for one) generates those questions again, every cell of them. When other ticked questions are waiting to be generated too, a second button such as Generate only the other 3 leaves the imported answers as they are and generates the rest.

Nothing else replaces imported answers in bulk. Generate all answers and Generate all answers in this section keep every cell that came from the file and fill only the empty ones, and Generate selected leaves alone an imported answer that someone approved, edited or sent back. The one exception is an answer marked Needs re read after its question changed, which every run writes again, keeping only the cells that came from the file, so an answer imported from the text of a Word document is written again whole. For one question, Replace with a generated answer on its card does it; on an answer imported from the text of a Word document, that button reads Refine.

First from your RFP knowledge. An answer your team has already approved can be reused from the Library for a matching question. A question steered by the project's Instructions for answering or an INSTRUCTION row is not given that answer as it stands: it is rewritten to follow them, and the rewrite is told to keep its facts, figures, names and claims, and when the approved answer does not hold what they ask for, the question is answered as if the Library had no answer for it. A question steered by guidance typed on its card, and every Refine, is written without the Library. Your RFP knowledge, which Zime searches for these answers, is kept for RFP answers alone: your product documentation, your security answers, and earlier approved RFP answers loaded into it; answers approved in Zime reach later questions through the Library, not through your RFP knowledge. It is searched apart from your general Zime knowledge, and every search is scoped to your company.

When your RFP knowledge holds a match for a question but not enough to answer it, Zime may also search your company's calls and shared documents, whatever the project's Knowledge scope. A question your RFP knowledge holds no match for is where Knowledge scope decides: with RFP knowledge, then general fallback, Zime writes a general answer for it instead, which can draw on your general Zime knowledge; with RFP knowledge only, it comes back marked Needs a source.

Each answer card lists what its answer rests on, under Sources: the documents its search returned, the Library for an answer reused from it, or the customer's file for an imported one, so you can check it against them before approving. A line under Sources names the instructions an answer was written with; an answer rewritten from an approved Library answer names that answer first under Sources, and its line starts "Rewritten from an approved Library answer".

Only where a question goes when your RFP knowledge has no match for it. It never changes what an answer counts: an answer reached through the wider setting counts exactly the same as any other. It is chosen as Knowledge scope on the New RFP project dialog, and a project owner or an RFP Admin changes it with the Knowledge button beside the project name on the Questions tab.

With RFP knowledge only, a question that neither the Library nor your RFP knowledge has a match for gets no general answer. It comes back marked Needs a source, and the Needs a source filter on the Questions tab shows just those questions, so the ones that want a person are never buried.

With RFP knowledge, then general fallback, those same questions get a general answer the next time they are generated, which can draw on your general Zime knowledge as well. Changing the setting does not answer anything by itself. Each general answer is labelled General answer on its card, so you can tell it apart from an answer drawn from your RFP knowledge. The label belongs to the version written as a general answer: an edit, a send back or a restore saves a new version without it. Refine writes a general answer only where the question may still get one, with this setting or after Release to general: a general answer is then refined as a general answer and keeps its label, and a Refine with no instruction of an answer a person, the Library or your file wrote gets a general answer when your RFP knowledge has no match for it. Otherwise, Refine never writes a general answer.

A single question can also be released on its own with Release to general on its card: its Needs a source marker clears right away, and the next time it is generated, it gets a general answer, unless the Library or your RFP knowledge has a match for it by then. The rest of the project keeps its setting.

Open the Settings tab in RFP, write your instructions under AI instructions and choose Save. They are given to the AI with every question of every project of your company, each time an answer is generated or refined. A project's Instructions for answering, and the instructions marked in its file, apply on top of them: where they disagree, generation is told to follow the more specific one. Only an RFP Admin can change them, on the Settings tab; everyone working on a project reads them under Instructions sent with this question. Your company has none until an RFP Admin saves some, and choosing Save with the box empty clears them.

When they are the only instructions, an answer approved in the Library is rewritten in your company's voice (we, our and us become your company's name) and checked against the approved answer; if it fails the check or cannot be rewritten, the approved answer is used word for word, marked Voice not applied. If a project's Instructions for answering or the instructions marked in its file also apply, the project's rule holds: a new answer is generated when the approved one cannot comply. The line under Sources of each answer written with them names your company's AI instructions.

Answers already written keep their words. The AI instructions reach an answer only when it is generated, so an existing answer takes a change only once it is generated again, such as with Refine.

You can edit the text yourself, refine it with a prompt, or regenerate it from scratch by refining with the prompt left empty. Nothing is overwritten: each of those saves a new version, the earlier ones stay on the card under history, each with its status, and Restore this version brings any earlier one back as a new copy, which counts no credits.

Editing, refining or restoring an approved answer takes the approval off it, because its text changed; approving again is what marks the new text ready for export. Choosing a different value in a dropdown on the card itself, without opening Edit, keeps the approval.

Zime found nothing to answer the question from, so it saved no answer and charged nothing. The question stays unanswered, and its card marks it No answer found, with the line "Zime found nothing to draw an answer from, so nothing was saved and nothing was charged. Add a document or write the answer." On a project whose Knowledge scope is RFP knowledge only, until the question is released, the line names your company's RFP knowledge instead and adds Release to general, which the card also offers beside Generate answer: choose it, and Zime can write a general answer for the question the next time it is generated. Where the question may get a general answer already, with RFP knowledge, then general fallback or after Release to general, Zime tried one before it marked the question No answer found.

It counts as unanswered, so the Unanswered chip on the Questions tab lists it, as it lists every question with no answer. It is not Needs a source, the mark of a question your RFP knowledge has no match for on a project that answers from it alone, so the Needs a source filter does not show it.

Generate all answers leaves such questions out, so running it again spends nothing on them, unless you tick Also retry questions with no answer found beside it, which is there while the project holds one. Generate all answers in this section leaves them out too, with no such choice. Generate selected asks again for every such question you tick, and so does Generate answer on its card. A run that leaves them out or meets new ones says how many, as in "Zime found no answer for 2 questions."

When Generate answer or Refine on a card gets nothing Zime can use, nothing is saved or charged, and the card says "Generate failed" or "Refine failed", with "Zime found nothing to answer this question from; nothing was saved and nothing was charged". After a Refine, the answer stays exactly as it was, its approval included.

Choose Approve on the card for one answer. For several, tick their cards and choose Approve selected; for a whole section, choose Approve all in this section in the section's three dots menu, on the Sections tab or in the list of sections on the left of the Questions tab; for several sections, tick them on the Sections tab and choose Approve selected sections; and for the whole project, choose Approve all on the Questions tab. Each of the bulk actions asks you to confirm first, and its message afterwards says how many answers it approved and how many it left, and why. Approve selected leaves out, before it asks, a ticked answer marked Needs re read, one that still has a part Zime found no source for and one in a frozen section, so its count and its message speak only of the answers it sends.

Under the assignment rules an Approve all approves only what you may approve. Answers waiting on a reviewer are left as they are, and the message names them: "12 answers need Alex's review and were skipped." The bulk actions also leave an answer marked Needs re read and an answer sent back that nobody has changed since; approve a re read answer on its card once you have read it. Nor do they approve an answer that still has a part Zime found no source for, which Approve on its card approves once you choose Approve with open gaps, or any answer in a frozen section.

An answer imported from the customer's file that still has an empty cell is not approved as it came in: Approve on its card names the cell, for example "Cell F12 is still empty. Fill it or generate it, then approve.", and the bulk actions leave it and say how many they left.

Approving an answer marks it ready for export. With the project's Approved answers enter the Library setting on, which is how every project starts, the answer is also added to the Library for an RFP Admin to review, and other RFPs reuse it only after an RFP Admin approves it; approving it again later refreshes that entry and keeps its status. With the setting off, the project's answers are never added to the Library.

Zime found no source for some parts of what the question asks. It wrote the answer from what it found and marked it Needs input, with how many such parts, such as Needs input (2): hover the mark to read them. Under the answer, Parts Zime found no source for lists each one, naming its cell on a row with several answer cells. Once you have dealt with a part, by editing the answer or writing in what it lacked for example, choose Resolve beside it: Resolve records your name and the time, the part leaves the list, and a line under Project Activity says so. Anyone who may edit the answer may resolve its parts.

Each version you make keeps the parts still open: an edit, a send back, and a restore, which brings back the parts still open on the version it restores. An answer Zime writes with Generate answer or Refine lists the parts it found no source for itself; one that starts from an approved Library answer, as it stands or rewritten to follow your instructions, also brings the parts that answer still had open, and one that keeps some cells of the answer before it, as Generate empty cells does, keeps the open parts of those cells.

Until each such part is resolved or the answer is approved, it counts as not answered: the Questions tab lists it under Unanswered and not under Answered, the Answered counts on the Questions tab, on the Sections tab and on the projects table leave it out, and its section reads In progress. Generate all answers does not write it again. In My Work it waits with its author, under To answer, marked Needs input, as the Track the work topic explains.

Approve on its card asks first, listing the parts still open: Approve with open gaps approves the answer as it is, and Cancel leaves it unapproved. The card then reads Approved, needs input, and the parts stay listed until they are resolved. Approve all, Approve all in this section and Approve selected sections leave such an answer unapproved and say how many, as in "2 answers still have parts Zime found no source for and were skipped.", and Approve selected leaves it out of the answers it approves.

An answer approved with open gaps goes into the file as approved. While the project holds one, the Export dialog says so, as in "Approved with open gaps. 2 approved answers still have parts Zime found no source for, and go into the file as approved:", and lists those questions, each a link to the question. With Include answers not yet approved on, in the same dialog, the export writes every answer as it is, so the dialog then lists every answer with such a part, approved or not, as in "Answers with open gaps. 2 answers still have parts Zime found no source for, and go into the file as they are:". On a project whose approved answers enter the Library, the Library takes such an answer with its open parts, so an answer that reuses it later shows Needs input with the same parts, and resolving a part of the approved answer clears it in the Library too.

A reviewer sends it back. Send back, beside Approve on the card, returns the answer to its author with an optional note saying what to change. The answer becomes a draft again, keeps its text so the author sees what they are being asked to change, and appears in the author's My Work under To answer with the note. A question with no author waits in nobody's To answer until one is assigned.

The note is shown to the question's reviewers and authors, the project owners and the RFP Admins. Anyone else on the project sees that the answer was sent back and by whom, but not the note. An answer already sent back cannot be sent back again until it has been changed or approved.

The Sections tab shows one of four words per section, worked out from its questions every time you look. Not started: nothing in the section has an answer yet. In progress: some answers exist, or all do but nobody has been asked to review them. Pending review: every question is answered, at least one is not approved, and a reviewer is set on the section or on one of those questions. Completed: every question's latest answer is approved. An answer imported from the customer's file counts as an answer here, even one with a cell still empty, so a section the file filled in can read In progress or Pending review before anything is generated. An answer not yet approved that has a part Zime found no source for is not counted as answered, here or on the projects table, until each such part is resolved or the answer is approved, so its section reads In progress until then. A frozen section keeps the status its answers give it; the padlock beside its name says it is frozen.

Answered counts the questions that have an answer, leaving out an answer not yet approved that still has a part Zime found no source for, the way Questions Answered counts them on the projects table, and Reviewed counts the approved answers, each against the questions that need one, subsections included.

Open the Sections tab and choose Freeze section in the three dots menu at the end of the section's row, or in the section's three dots menu in the list of sections on the left of the Questions tab. Zime asks first, with the section's progress and its authors and reviewers: "While Security is frozen, nobody can answer, edit, approve, send back, assign or change its questions, the person who froze it included, until it is unfrozen." Choose Freeze section there, and the section is frozen with its subsections. To freeze several at once, tick them on the Sections tab, choose Freeze selected sections on the bar above the table, and confirm with Freeze 3 sections, for three. A project owner, an RFP Admin and the authors and reviewers its row on the Sections tab shows can freeze and unfreeze it; someone assigned only to some of its questions cannot. While a Generate run is going, Freeze is refused, so pause the run first.

While a section is frozen, Zime refuses every change to it and to its questions, whoever asks, the person who froze it, project owners and RFP Admins included: nobody can generate, write, edit, refine, restore, approve or send back an answer there, resolve a part Zime found no source for, change who is assigned to it, attach or remove a file, or add, edit, move or delete a question or change its answer type. Generate all answers and Approve all go on with every other section and say how many they left as they are, as in "3 in frozen sections were left as they are." The section's approved answers still go into the export, its status on the Sections tab stays as its answers make it, and its questions stay in My Work. An answer Zime was still writing for one of its questions when it was frozen is not saved, and nothing is charged for it.

A padlock beside the section's name, on the Sections tab and in the list of sections on the Questions tab, shows it is frozen: hover it to read who froze it and when, such as "Frozen by Alex on 19th Oct 2026 9AM". On the Questions tab its questions open with a line such as "Security is frozen by Alex since 19th Oct 2026 9AM. Unfreeze it to make changes.", whose second sentence only someone who may unfreeze it sees, and each of its cards carries the same line. Every menu of the section and of its cards opens with a line saying why, and the controls the freeze turns off stay in view, off, for the people who could use them once it is unfrozen; on the Sections tab a dash takes the place of the plus beside its Authors and Reviewers, with the line as its hover text.

On a project that is not archived or in Trash, Generate selected and Approve selected, on the bar above the questions, leave out the ticked questions of a frozen section, and the bar says how many, as in "1 is in a frozen section and is left out"; Users and Edit Answer Type are off while one is ticked, with the line under the bar. When every question of such a project is in a frozen section, Generate all answers and Approve all are off, and the line under them says "Every question of this project is in a frozen section." On the Sections tab, Approve selected sections leaves a ticked frozen section out, and its confirm says so: "1 ticked section is frozen and is left out."

Unfreeze section, in the same menus, and Unfreeze selected sections on the bar unfreeze it, for the same people, after a confirm that says everyone who may answer, edit, approve or assign in it can do so again. A subsection frozen with its section is unfrozen with it: on the subsection itself Unfreeze section is off, and its menu names the section to unfreeze. A subsection frozen on its own before its section was frozen keeps that freeze: unfreezing the section leaves it frozen, as the confirm says, and unfreezing the subsection while its section is still frozen lifts only its own freeze, which the confirm says too: "It stays frozen through Security until Security is unfrozen." Each freeze and unfreeze is a line under Project Activity.

No question is added to or moved into a frozen section: Add question is off on it, and the section lists of Add question and Move to section offer only the sections that are open. When Apply changes would change a frozen section, its preview names the section, who froze it and since when, and says "This apply would change it. Unfreeze it to apply.", and Apply stays off until it is unfrozen. Rebuild from file is refused while any section of the project is frozen, and so is a Restore from Removed by Apply that would bring a question back into one.

On the projects table, in the Project Due Date column, and at the top of the project, beside its name. Both show the due date and time set on the project, in your own time zone.

Open My Work, beside Projects. To answer, at the top, counts every question you are the author of that has no answer yet, was sent back to you, holds an answer imported from the customer's file that still has an empty cell to fill, or holds an answer, not yet approved, with a part Zime found no source for. To review counts every answered question you are the reviewer of that is neither approved nor sent back, apart from two kinds of answer that wait in their author's To answer instead: such an imported answer, until that cell is filled or the answer is changed, and an answer with a part Zime found no source for, until each such part is resolved or the answer is approved; on a question with no author, an answer of that second kind stays in its reviewer's To review. A project you only own does not appear here; My Work follows the questions you were made an author or a reviewer of, and lists the ones still waiting on you.

A question you answer leaves To answer; if a reviewer is named on it, it waits under their To review until it is approved, and it returns to your To answer list if they send it back. Nothing was lost and nothing was unassigned: the question is still yours, as its card and the Team tab show, it is simply no longer waiting on you. A send back carries the reviewer's note when they leave one; the note is optional and a send back without one says only that it came back.

Under the two counts the questions wait in one block per project, which says how many wait for your answer and how many for your review, as in "2 to answer · 1 to review", and when the soonest of them is due, written the way the projects table writes a due date. Blocks with a question past its due date come first, marked Overdue in red, then the others, soonest due first, then those with no date. Inside a block the questions follow the customer's file, each marked Not answered, Sent back, Needs input or Pending review, with the reviewer's note when they left one. Needs input marks an answer with a part Zime found no source for: open the question, and once you have dealt with a part, choose Resolve beside it under Parts Zime found no source for, which records your name and the time; when every part is resolved, the answer counts as answered and moves on as any answer does. When your assignment was given a due date of its own, its row shows that date, and Overdue and the date filters follow it; otherwise the row is due when its project is. A question in an archived project is never overdue, and its row says "Archived, read only: unarchive the project to finish it".

Choose a question's label, at the top of its row, to open it: its project opens on the Questions tab, in the question's section, with the question in view and My questions on, so a section shows only the questions you are an author or a reviewer of. Choose My questions again, or All, to see every question.

Narrow the page with the filters above the counts: Due date (Next 3 days, Next 7 days, Next 30 days, Next 90 days or All), Task type (To answer, To review or All), Project, and Overdue only. A date window keeps what is already late too, and a question with no date shows under All only; when nothing matches, the page says "No question matches these filters." The two counts at the top always count everything waiting on you. Your filters are kept in this browser, for you alone, and are still set when you come back to My Work; a browser that blocks what sites keep holds them only until you leave the page.

An RFP Admin also sees, while answers wait for their approval in the Library, a line saying how many, such as "Library: 12 answers wait for your approval"; choose it to open those answers.

To review lists answered questions only. Being assigned puts nothing there on its own: the question appears once it has an answer, whether that answer was generated, typed, reused from the Library or imported from the customer's file. An imported answer that still has an empty cell waits in its author's To answer instead, until that cell is filled or the answer is changed. So does an answer, not yet approved, with a part Zime found no source for, marked Needs input, until each such part is resolved or the answer is approved; on a question with no author it stays in your To review. Until a question has an answer it sits in its author's To answer list, if it has an author, not in yours.

If My Work was already open when the answers were generated, it still showed the list as it was when you opened it. It reads the list again whenever you come back to its browser tab or window, and Refresh at the top reads it again at any time.

On the Team tab, which lists the owners, the authors and the reviewers of the project, each once, with the roles they hold and what they are assigned to. Add takes you to the Sections tab, where people are assigned. Remove, which a project owner or an RFP Admin sees beside anyone with an assignment, takes that person off every section and question they were assigned to; anyone who is not an owner then leaves the project, though an RFP Admin still sees every project. When a section they are on has questions with people of their own, Remove is refused and nothing changes: take them off that section with Manage Users on the Sections tab, which asks whether to reset those questions. Owners are changed on the About form.

Open the Activities tab. Project Activity is the project's own feed, newest first: its creation, the import and every Apply changes, each Restore from Removed by Apply, with any section it brought back, and each Delete Permanently there, every change to the About form and the owners, every assignment and every due date given in Manage Users, with who gave it to whom, every Edit Answer Type save, each answer written on a card (generated, reused, refined, typed, edited or restored), one line per Generate run with its count and how long it took, each approval, single or in bulk, each time approvals were taken back to draft and each time they were restored, each send back, each export, saying how many answers not yet approved it included when there were any, and the archive and trash moves. Where your company's RFP emails are on, each Send and each Delete of queued notifications is a line too, while Send email notifications for this project, in the About form, writes none. Filter it by person or by kind, search the lines loaded so far, and Load more brings older lines. Everyone on the project sees it. A send back note is never in it.

Q&A Activity lists each question's latest answer: the kind of answer it is and who approved it or sent it back, with a link to the question. Earlier versions are not listed there; they stay on the card under history.

Yes. Configuration, beside Export at the top of the Questions tab, opens the review screen again with your marks and the grid exactly as you left them; a project owner or an RFP Admin sees it. Mark what needs to change, then choose Apply changes.

Apply changes reads your marks against the project one question at a time. A question whose wording, answer type, dropdown values or section changed is reset: it stays the same question, with the people assigned to it and its answer text, at no additional credit. An answer that was approved or edited drops to draft and shows Needs re read, so you know to read it again against the new question; an answer that was still a draft stays exactly as it was. A question you did not change keeps its answer and its approval. A row that is new becomes a new question, and a heading that is new becomes a new section. A question with no people of its own answers to its section's, so a question that moves to another section can answer to different people.

A row that is no longer a question goes to Removed by Apply on the Sections tab, listed under the section it sat in and kept with its answers, people and files. There a project owner or an RFP Admin can Restore it, which brings it back with its answers and marks its row again so the next Apply keeps it, or Delete Permanently, which takes it off the list for good. If its row asks something different by then, an answer that was approved or edited comes back as a draft marked Needs re read. A section left with no heading, no question, no sub section and nobody assigned to it is removed, and in a Word document a heading Zime found counts only while a question sits under it, unless you marked it yourself. Restoring the questions of a section an Apply removed brings that section back too, under its own heading and in its own place.

Choosing Apply changes writes nothing by itself: it shows you a preview of all of that first, and nothing changes until you choose Apply there. It cannot run while a Generate run is going, so pause the run first. When it is done, the project opens on its Sections tab.

Rebuild from file, under More on the same screen, is the older way: it applies every mark and removes every answer, every approval and every assignment in the project, then imports the answers the customer's file holds again, as the first import did. Questions added by hand go too, every change made on a card with Edit question, Move to section or Delete is undone, and the files attached to questions are removed. It asks you to confirm first and says how many answers and assignments would go, and the credits already spent on them. It reads the file you already uploaded, so it is not a new upload. It is not the way to mark a sheet again from scratch. For that, choose Clear Sheet on the bar above the sheet, then Clear to a blank sheet, mark the rows, then choose Apply changes.

The whole change, in the order you need it: the new sections it would add, by name; the sections it renames or moves in the tree, and how many questions then answer to other people; how many new questions it would add; the questions it would reset, with their names and how many of those were approved, and how many answers drop to draft for you to re read (their text is kept, at no additional credit); what it does to the answer cells, such as a column added or taken off and the cells it takes from the customer's file; the rows it would remove, by name; the empty sections it removes; and how many answers stay exactly as they are.

When the change drops every answer in the project to draft for you to re read, the preview says so in capitals. A reset of every answer can never happen quietly, and the answer text still stays: only its status drops.

A removed row keeps its answers: the preview says the removed rows go to Removed by Apply on the Sections tab with their answers, people and files, and that Restore puts each one back. The line that lists them says how many answers go with them, how many of those were approved and the credits already spent on them, and when every answer in the project goes there, the preview says so in capitals as well.

Back closes the preview and changes nothing. Apply carries it out. If the project changed while the preview was open in a way that changes what the preview says, Apply writes nothing and shows you the current preview instead, so what you accept is what you get.

Credits already spent on an answer that is reset or removed stay on your meter, so a month's total never quietly falls, and a Restore charges nothing. Apply changes emails nobody when it runs. A reset answer that was approved or edited keeps its text, shows Needs re read on its card and, where it has a reviewer, waits in My Work under their To review, not in the author's To answer, so where your company's RFP emails are on it can come up in that reviewer's Daily Task report like the rest of their To review; a reset answer that was still a draft stays exactly as it was. The Activities tab carries a line naming what the apply changed, and one for each Restore.

Reset: the wording of a question, its answer type, the values in its dropdown or their order, and the section it belongs to. A question changes section when you mark a heading above it or take one away, so a heading marked over answered questions resets the questions under it. A new Dropdown or Scale answer cell resets the question too, because it asks for something the answer never gave, and Edit Answer Type resets by the same rule. Reset here means that an answer that was approved or edited drops to draft and shows Needs re read, for you to read again against the new question, at no additional credit, while an answer that was still a draft stays a draft with no mark. Either way the answer itself stays exactly as it was.

Kept: where the answer is written. Moving the answer cell, or correcting the wording of a heading without moving it, changes nothing about what is being asked, so those answers stay approved, and so does a changed character or word limit or Value in File (what the export writes for a dropdown value), or a text answer cell added or taken off. A dropdown value chosen in Zime moves with its cell, and is written in its new cell even where the customer filled that cell; a value your customer's file gave stays in the customer's own cell. The Library keeps each value in the column it was chosen for, so a later project that reuses the answer gets it there. Cells of an added answer column that the customer's file fills come in as the customer wrote them, and the preview says which empty new cells the next Generate all answers fills and which stay empty because their answer is approved or sent back. If a change leaves an answer nothing to write into, its status drops to draft too, and the preview counts it with the resets.

Choosing a different header row, or moving the Question column or an answer column, is read the same way: every question is read again, and only the questions that now ask something different are reset. A row that ends up at or above the header row stops being a question and goes to Removed by Apply.

A question you edited on its card keeps your edits, unless the changes you are applying also touch that same row on the review screen, in which case the review screen wins.

Choose Select by Requested Answer on the bar above the questions on the Questions tab, then choose a value. It lists each value, spelled as the file spells it, with the number of questions it ticks. Two values that differ only in capitals are listed apart, each with its own count, so choose each spelling you mean. A question's value, shown as a tag on its card, is the kind of answer its row asks for: its cell in the customer's Requested Answer column, where Zime reads one, or, on a row a rule in Rules by value covers, its value in the rule's column.

Choosing a value ticks every question that carries it, in every section, and adds them to what you have ticked already, so you can choose a second value or tick more by hand. Clear, on the same bar, empties the selection. It ticks only what you could tick by hand: the questions the filter in use shows, never one whose card reads Information only.

Then choose More on the same bar and Edit Answer Type to change the answer columns of all of them at once, without typing a row number: delete a comment column there, for example, to take that cell off every one of them. Save first shows you what it will change, and nothing is written until you choose Save there; the dialog also says how many of the ticked questions it cannot change, and why. Anyone who opens the project can tick questions this way; More, with Edit Answer Type, is there for a project owner or an RFP Admin.

Read the questions first, right after the import, before anything is generated or approved: fix the wording, the sections, the header row and the answer cells on the review screen, then Apply changes. Every change applies cleanly at that point, because nothing generated or approved is there to reset.

Then generate, review and approve. A change that resets a question whose answer was approved or edited drops that answer to draft and marks it Needs re read, at no additional credit: the answer text is kept exactly as it was, and it is the approval that is asked for again. Approve each such answer on its card once you have read it: the bulk approvals skip them, and a Generate run would write them again, which counts like any newly generated answer, keeping any cell that came from the customer's file and leaving alone one that is still wholly the file's. A reset answer that was still a draft gets no mark, and Approve all approves it like any other draft, so read the questions the preview names as reset before you approve. The preview puts all of that in front of you before anything happens, and Back is always there.

A finished project can be reopened the same way. Completed is worked out from the approvals rather than set by anyone, so a project drops back to Active the moment one answer drops to draft on a reset row, and returns to Completed once every question's answer is approved again. Export keeps working all the while: it writes only the answers approved at that moment, unless you turn on Include answers not yet approved in the Export dialog, which writes the latest answer of every question as it is.

Choose Export at the end of the project's tabs, after Activities, on any tab, or Export at the top of the Questions tab. Both open the same Export dialog, and only a project owner or an RFP Admin sees them. On Sections, Team or Activities, Export takes you to the Questions tab and opens the dialog there. When the project cannot be exported yet, for example while no question has an answer, Export tells you why: on the Questions tab, hover Export at the end of the tabs; on any other tab, choose it, and the Questions tab opens with "Export is not available" and the reason.

One file, and which kind depends on what you uploaded. When an answer in it has attachments or images and Attachments to Include in the Export dialog is left on All, the download is a zip holding that file with the attachments and images in folders beside it, and each answer with attachments ends with a line naming them. Choose None to get the file alone.

An Excel workbook comes back as your own file with the approved answers written into its answer cells (.xlsx, the macro enabled .xlsm included, macros intact). A Word document (.docx) comes back as your own document with each approved answer under its question, or in its cell when the question sits in a table. Where the document already held an answer that you marked ANSWER on the review screen, it stays as it was unless you approve a changed answer, which then takes its place.

A Word project can also export a Table: choose Table instead of Source (.docx) under Export Type, and you get a Word document of its own holding one table, a row per clause of your customer's document with its number, its compliance status marked in its column, and its approved answer. Your customer's document itself is not written into.

A PDF questionnaire comes back as one Word document (.docx); the original file is never edited, and no PDF is made. With Questions and answers only off, which is how the Export dialog opens, it is your customer's whole PDF rebuilt in Word: every line in its own font, size and colour, each drawn table as a Word table with its own column widths and shading, a table that runs over two pages as one table, the pictures, and the running header and footer as the document's own. Each approved answer is written in at its question: into its answer cell when the question is a row of a table of questions, into the answer row under a question drawn as a box, in place of what that row printed, inside its cell after the question's last line when the question is printed in another table, and otherwise on a new line under the question, after any dots that end its line. A row an answer goes into grows with it. An approved answer whose question was added on the project has no place in the PDF: it goes at the end of the document under "Additional questions and answers", and the Export dialog and the line the export writes under Project Activity say so.

With Questions and answers only on, it holds only the questions and their approved answers, in your customer's formatting: each table that came in as a table of questions under the heading above it, with every row as your customer printed it and each approved answer in its own answer cells, then the other approved questions with their answers in their places; a PDF read as one table comes back as that table, with a Response column when no column takes the answers; any other PDF as its approved questions and answers, section by section. Headings keep their numbers and colours, and tables their header colours, column widths, shading and rules. Nothing Zime set aside, no running header or footer and no picture is in it. When the original file is not available, or its formatting cannot be read within the time and memory an export may use, the document comes in plain formatting, and the line the export writes under Project Activity says so.

When the whole document cannot be built, because the PDF has more pages or is larger than an export rebuilds, its rebuild would take longer or need more memory than an export may use, a page of it is rotated or holds text written right to left, or an approved answer of the file has no place in it, nothing is downloaded: the Export dialog says why under Questions and answers only and offers it, and with it on the export still goes through. When other PDF files are being read or exported at that moment, the dialog says so in the same place: try again in a few minutes, or turn Questions and answers only on. The whole document needs your customer's original file: when it is no longer available, the project shows Attach original file, which takes back that same file, and Questions and answers only exports without it. To send a PDF, open the Word document in Word and save it as PDF.

Only approved answers are exported unless you turn on Include answers not yet approved in the Export dialog, which is off whenever the dialog opens. Off, anything still in draft is left out, so an answer reaches your customer only after a person has agreed to it. On, the latest answer of every question is exported, approved or not, each in the place this page names for an approved answer, and an answer not yet approved is written as it is. In a workbook or a Word document, leaving a draft out never removes what the customer wrote: the export starts from their file, so text they wrote in it stays wherever no approved answer is written. So does the text of a PDF in its whole document, with Questions and answers only off.

Exporting counts no credits, whatever comes back, the zip with attachments included, and however often you export a project.

Because only approved answers are exported while Include answers not yet approved is off in the Export dialog, which is how it opens, and most of yours are probably still drafts. Generate all answers writes drafts, and with it off nothing reaches the file until a person approves it. One approved answer is enough for such an export, so a file with one filled cell and hundreds of drafts behind it is that rule working, not a broken export. While drafts remain and Include answers not yet approved stays off, the Export dialog says how many of the answered questions are approved and that the drafts will not be in the file, and offers Approve all first; with no answer approved at all it keeps Export off and says "No approved answers yet. Approve answers, or turn on Include answers not yet approved to export the answers as they are." A draft that still holds exactly what your customer's file gave it is not counted there: in a workbook or a Word document, and in the whole document of a PDF while Questions and answers only is off, the file keeps it as they wrote it.

The Approved chip on the Questions tab shows how many answers are approved. Into a workbook an approved answer is written only where the file gives it an answer cell: a question you added by hand has none until a project owner or an RFP Admin types its cell in Answer text goes to, in Edit question on its card, and until then its answer stays in the project and out of the file, as its card and the Export dialog both say. A question added by hand to a Word project has no place in the document, so its answer stays in the project. The Word document of a PDF carries every approved answer, the answer of a question added by hand included, except for text in a table whose only answer column is a Dropdown or a Scale, which a Word document leaves out too: in such a table, in a Word document or in a PDF whose tables came in as tables of questions, the export writes each approved choice into that column and leaves out any text the same answer holds, since the table has no column to write it in. To have that text exported, write it in a text Answer column of the table, giving the table one in Table Properties on the review screen if it has none.

To include the drafts, approve them (or choose Approve all) and export again. To send them before anyone has approved them, turn on Include answers not yet approved in the Export dialog and export again: every question then gets its latest answer, written as it is, and the dialog says how many of them are not yet approved. Either way the export always starts from the file the customer sent, so nothing needs undoing. Exporting again counts no credits.

Yes, in a workbook and in a Word document. Zime never re saves your customer's file through spreadsheet or word processing software. It opens the original, writes the answers in, and copies every part it does not need to change byte for byte, in the same order, with the same compression.

In a workbook only the answer cells change, so workbook and sheet protection, named ranges, dropdown lists, conditional formats, comments, external links, custom XML, sensitivity labels, row heights and every style come through intact, and sheets Zime did not answer are untouched. Three small changes can go with the answers: a workbook that holds formulas is set to recalculate them when it opens, Custom Font, a font choice in the Export dialog, adds its font to the file's styles, and a link on a cell an answer replaces goes with the text it linked. Formula cells are never written over, and neither is any other cell of a column your customer's formulas fill on the question rows; the review screen names such cells before you export.

In a Word document each answer goes in under its question as paragraphs of its own, keeping its bold, italics, links, lists and images, in the question's font unless you choose Custom Font and never in its colour; or into its table cell; or in place of an answer you marked ANSWER. An answer never takes the question's numbering, so it never reads as another clause of the document, and your customer's own styles, headers and footers are left as they are. Track changes and protection stay exactly as the customer set them, and the answers go in directly, not as tracked changes; the Export dialog says so for a document that has either. A field, such as a table of contents or a page number, is never rewritten.

Every export into your customer's own file is checked before it reaches you. Zime re opens the file it has just produced and checks that every part it did not change is identical to the original and that each answer sits where it was meant to, and in a workbook also that nothing else on a written row changed and that every formula cell is still there. If that check fails the file is not handed over and you are told why. The Word document of a PDF is checked the same way against the document Zime rebuilt from the PDF: the check says each answer sits where it was meant to, and not how closely the rebuilt document follows the look of the PDF. A Word Table export is a new file, so it is not checked this way.

A PDF comes back as a Word document rebuilt from it: each line in the font, size and colour the PDF gives it, each table with its column widths, shading and rules, and a new page where each page of the PDF begins, except inside a table that runs on. Word lays the text out again, so a page may not end exactly where the page of the PDF did, and an answer moves what follows it down.

Not into the PDF file itself, and the export is not a PDF: a PDF questionnaire comes back as a Word document (.docx), which you can edit, and which you can open in Word and save as PDF when a PDF is what you send. Export Type in the Export dialog reads Word document (.docx), and Font Format applies to it as to a Word project: Source Font (Normal Style) writes each answer in the font and size of its place, and Custom Font in the font you choose.

Questions and answers only, in the Export dialog, decides what the document holds. Off, which is how the dialog opens, it is your customer's whole PDF rebuilt in Word, each approved answer written in at its question: a questionnaire that draws each question as a small table with an answer row under it gets the answer in that row, in place of what the row held; a question in a table of questions gets it in its answer cell; a question printed in any other table gets it inside its cell, after the question's last line; any other question gets it on a new line under the question, after any dots that end its line. A row an answer goes into grows with it, and what follows moves down. On, it holds only the questions and their approved answers, in your customer's formatting. With Include answers not yet approved on, in the same dialog, the latest answer of every question goes where an approved answer goes, approved or not.

An answer keeps every character, such as ś or Ł and any Cyrillic, Greek or Chinese character, and its bold, italics, lists, links and images.

The whole document is refused, with the reason, when the PDF has more pages or is larger than an export rebuilds, when its rebuild would take longer or need more memory than an export may use, when a page of it is rotated or holds text written right to left, or when an approved answer of the file has no place in it; the Export dialog then offers Questions and answers only, which still exports.

Zime does not fill in the fields of a fillable PDF form. If the answers have to go into the document itself, convert the PDF to Microsoft Word first and upload the .docx.

Reuse. Each project decides whether its approved answers enter the Library, with Approved answers enter the Library on the New RFP project dialog and on the About form. Left on, which is how every project starts, approving an answer adds it to the Library's Pending tab, and no other RFP reuses it until an RFP Admin approves it there. Turned off, the project's answers are never added to the Library.

Only approved entries that are not stale are ever reused on a later question. A stale badge on a row means that entry has been taken out of reuse, and it is never served again while the badge is there, whatever its status. An RFP Admin can approve or reject an entry on the Library page while it is still pending, and that decision is never overridden: approving the same answer again refreshes the entry's question and answer, every cell included, rather than creating a duplicate, and leaves the entry where the Admin put it. A rejected entry stays rejected, and an entry promoted from Pending goes on being reused unless it is marked stale. To decide several at once, tick their rows on the Pending tab, or tick Select all for every row on the page, then choose Approve selected or Reject selected and confirm; when more entries wait than the page shows, the link that appears beside Select all selects every pending entry, or every one your search finds. In an entry, a dropdown or score value stays under the column it was chosen for, even after it is moved to another cell. An entry made from an answer approved with parts Zime found no source for shows Needs input with how many; hover it to read them. An answer that reuses such an entry carries those parts, and while the project's setting is on, resolving a part of the approved answer clears it in the entry too.

Reuse is also held back where it would not fit. An entry approved on a project with a named answer language (Answer language, beside the project name) is not reused on a project that names a different one. A question steered by its project's Instructions for answering or an INSTRUCTION row is not given an approved answer as it stands: it is rewritten to follow them, and when it does not hold what they ask for, the question is answered as if the Library had no answer for it. One steered by guidance you typed is written without the Library.

The setting decides what each approval does from the moment you change it. Entries already in the Library stay where they are, so turning it off is not a withdrawal of what the project has already contributed, and while the setting is off, approving an answer again does not change its entry.

Per answer Zime writes or reuses, at one flat rate, and per file you upload: each upload uses 2 credits.

An answer served from the Library counts exactly the same as one written from scratch.

An upload is the file you upload to start a project, the questionnaire your customer sent, and a new version of that questionnaire is a new upload. It counts once the project is created from it, before you review what Zime read. Apply changes and Rebuild from file read the file you already uploaded, and Attach original file takes back that same file, so none of them is an upload, and neither is a file you attach to an answer.

Each new version Zime writes counts, even on a question that already has an answer: a refine or a regenerate, Generate empty cells, Generate selected when you choose Replace them (Replace it, for one) or when it writes again an answer marked Needs re read, and a run of Generate all answers or Generate all answers in this section that fills cells an answer left empty or writes again an answer marked Needs re read. Editing the text yourself, writing an answer by hand, restoring an earlier version, the answers your customer's file came with, approving, sending back, exporting and archiving count nothing at all.

A question that ends with no answer is never counted. If neither the Library nor your RFP knowledge has a match for it, the project's Knowledge scope is RFP knowledge only and the question was not released with Release to general, it is marked Needs a source and nothing is charged for it.

An upload that fails costs nothing: it counts only once the project is created from the file.

Open Projects. The Credits column of the projects table shows each project's credits in total, and the table can be sorted by it; it lists only the projects you can open. Inside a project, This RFP, beside its name, shows the credits it used this month. Each of these credit figures includes the uploads.

Choose Credits and usage on Projects to open the page that shows the AMA credits you used this month and lists each charge, your RFP answers and uploads among them. It is there when your company has Ask Zime and a plan that counts AMA credits.

Every counted answer is recorded once, at the moment the answer is saved, with its project, its question, what it counted and the month it belongs to. The answer and its record are written together in one step, so a counted answer always carries exactly one record and an answer that failed to save carries none. Every upload is recorded the same way, once, at the moment it succeeds, with its project, what it counted and the month it belongs to, and one that failed carries none.

Each of these figures is the sum of those records, so every credit in it traces back to one saved answer or one upload. A project removed for good keeps its records, though no row of the table shows it any more.

Months are whole calendar months in UTC, so an answer or an upload never lands in two of them.

Yes. An RFP answer or an upload is charged to the person who asked for it, the same way as their questions to Ask Zime: first to their own monthly AMA credits, then, if your plan has a company pool, to that pool, up to their monthly credit limit. Their RFP answers and uploads and their questions to Ask Zime count toward the same monthly limit.

Whoever chooses Generate answer, Generate empty cells or Refine on a card asks for that answer, whoever starts Generate all answers, Generate selected or Generate all answers in this section asks for the answers of that run, and whoever chooses Upload and map asks for that upload.

Only what counts RFP credits is charged to your AMA credits, each at the same count as its RFP credits: the answers Zime writes or reuses from the Library, through Generate answer, Generate empty cells, Refine, Generate all answers, Generate selected or Generate all answers in this section, and each upload, at 2 credits. What counts no RFP credits, such as an export, is never charged there either.

The charge is made up front. Before Zime starts an answer or reads a file you upload, it sets the credits aside from your AMA credits, so one they cannot cover is never started. If it then fails, the credits are given back, so one that fails costs nothing.

When the AMA credits an answer or an upload would be charged to cannot cover it, Zime does not start it, and nothing is written or charged for it. The screen gives one of two reasons, in these words.

Once the person who asked for it has reached their monthly credit limit: "You have reached your monthly credit limit. It resets on the 1st." The limit is that person's alone, so it stops nobody else.

Once your company's pool cannot cover it: "Your company is out of credits. Contact Customer Success to top up." Meanwhile, anyone whose own monthly AMA credits still cover an answer or an upload can go on generating and uploading.

On a card, those words come with "Generate failed", "Generate empty cells failed" or "Refine failed", and the question keeps the answer it had, if it had one.

An upload is refused in the New RFP project dialog with "Upload failed" and those words: no project is created, and the dialog keeps what you filled in. An export counts no credits, so it is never refused for them.

Generate all answers, Generate selected and Generate all answers in this section stop at the first answer the credits cannot cover. Answers already under way are still saved, and the questions the run had not reached stay as they were. The Questions tab then shows "Batch stopped at 40/120." for a run with an answer on 40 of its 120 questions, followed by the same words, and the run's line under Project Activity says it stopped and why: "monthly credit limit reached" or "out of credits".

Nothing else stops. Every draft and every approved answer stays as it is, and editing, writing an answer by hand, approving, sending back and exporting go on working, as none of them counts credits.

Once the limit resets on the 1st, or Customer Success has topped up the pool, run it again: Generate all answers picks up the questions still waiting across the whole project, Generate all answers in this section those of that section, and Generate selected the questions you tick. For a file, choose Upload and map in the New RFP project dialog again.

It is copied to Zime storage before the project is created. If that copy cannot be made the import is refused and the upload costs nothing, so a project always has a durable copy of the file it was built from.

Exports and re reads of a sheet work from that copy, so maintenance on Zime's side cannot lose your file.

The copy stays for as long as the project does. Archiving the project or moving it to trash leaves it in place, and nothing removes it before the project itself is removed for good.

The file a project was built from is never replaced. Attach original file, which a project shows only when it has no usable copy, accepts only that same file: a file whose sheets, headers, columns, counts or questions differ from what was imported is refused, so answers can never be written into a different file from the one they were mapped to. A new version of the questionnaire comes in as a project of its own, from New Project, and counts as an upload.

A copy is removed only when its project is removed for good.

Choose Archive in the row menu on the projects table. An archived project is closed for work, not only for spending: the only actions still open on it are Export, Unarchive and Move to trash. Everything else is refused until you unarchive, so editing an answer, approving one, assigning people, generating and re importing all come back as a refusal. It stays readable and exportable under the Archived tab, and Unarchive brings it back at any time.

Move to trash is for a project you no longer need. It sits under the Trash tab for 30 days, where Restore brings it back untouched; after that it is removed for good. Both are for a project owner or an RFP Admin, and neither works while a Generate run is going, so pause it first. Answers it put in the Library stay there, and the approved ones go on being reused, even once the project is removed for good.