| Before Zime | After Zime | Solutions with Zime | |
|---|---|---|---|
| Right judgment | Winning behaviors never written down; guidance too generic to act on. | Behaviors matched to motion, role, product line, and competitor: 12% more technical wins. | Behavior Graph: Zime interviews with Versa's SE and revenue leadership, and Versa's own call data. Delivered stage-by-stage discovery and POC playbooks. |
| Adoption | Leaders couldn't see inside SE calls; org-wide dashboards showed everything, so they showed nothing. | 79% of engineers running the right technical discovery in three months. | Stage-gated checklists, context-carrying prep notes, objection-flagged pipeline reviews, scheduled reports, delivered in Teams and Salesforce. |
| Production harness | Battlecards, role-play content and handover notes maintained by hand, going stale as products and competitors changed. | Judgment kept current and governed as products, competitors and strategy change. | Scoped visibility by region and role, versioned playbooks, competitor entity resolution, handover and RFP skills, per-project credit and cost reporting. |
Versa sells multiple product lines against ten named competitors, in deals from $100K to $3M. Almost every deal turns on a technical evaluation. In that evaluation, an engineer has to establish four things:
What is the current tech stack?
What is actually broken?
What are they moving to?
Does the proposed POC have real test cases, timelines, and budget behind it?
Get it wrong, and the deal sits in the forecast looking healthy for two quarters before it dies. Versa's engineers did that work well. The problem was that it happened once, in one call, and then vanished.
Writing it up was slow, so most of the details never reached Salesforce. Nor could managers simply sit in. John Volkering, who ran the European SE team, explained why.
So the only way to know how a deep dive had gone was to ask afterward. Mickey Singh, who runs sales enablement, named the gap: you have no idea what's happening in your SE calls, or how the technical content was presented.
The expensive failure was the AE-to-SE handover. By the time a deal reached an SE, the BDR and the AE had already established budget, authority, champion, and pain. None of it traveled, so the SE asked again. It broke five ways.
Unclear requirement. The engineer walked in not knowing what to test. Fit, or no fit, got settled on the call meant to prove it.
Looking for guidance. When an engineer hit an objection or a competitor they hadn't prepped for, the only option was to guess, or stop the call to find someone who might not be free.
Mangled acronyms. Product names and protocols have to be captured exactly. Get one wrong, and the record breaks at the very points a technical call turns on.
Broken telephone. What reached the engineer was the AE's retelling. Two versions later, it was a different requirement, and the POC got built for that.
Not salespeople. Engineers have no interest in becoming salespeople.
Mickey Singh was blunt about the last one.
Left alone, the technical call goes deep on features while the commercial thread goes quiet. All of this surfaced at the top as forecast noise. Here is Martin Mackay, Versa's CRO, on what his pipeline held:
Versa was not a late adopter waiting to be sold to. Chitresh was already putting AI into Versa's revenue motion before most of its market was.
He had more than one AI tool reading his deals, and when two disagreed about an account's health, he ordered a forensic review of both rather than taking the one he preferred. He was specific about what to build: feed it Versa's competitor list and battlecards, so it hears Fortinet or Zscaler named and tells an engineer whether the rebuttal held; draft the follow-up from what the customer actually said; score the sentiment and name the detractors in the room.
He was just as precise about where the build stopped. In-house agents gave him visibility, not adoption across 200+ reps, and closing that gap was a 12-24-month build across multiple use cases: constant iteration, tribal knowledge encoded into hundreds of skills, and all of it kept current as the competitive set moved. Rather than spend that time and a million dollars in GTM engineering to build it himself, he chose to get the same judgment running with Zime, at a fraction of the cost. Building that judgment and getting the field to run it, not just seeing it on a dashboard, is what Zime delivered with the Behavior Graph.
Three pieces of work, in this order. Encode the judgment Versa's best SEs, AEs, and BDRs already had. Get it into the calls where it decides something. Then keep it true as the products and the competitive set move.
Zime built a Behavior Graph for Versa by encoding:
SE expertise. Carried in from technical-selling organizations that had solved the same problem. E.g., how top SEs do technical discovery, stick to ROI use cases rather than feature parades.
Elicited judgment and strategy. Interviews with Chitresh and Versa's enablement and revenue leaders on what good looks like in a Versa technical evaluation, including what never shows up in a transcript.
The CRO's strategy. Martin Mackay's forward-looking priority: move early-stage deals through or qualify them out, and reinforce outcome-centric selling in live calls rather than in a training deck.
Versa's own data. Used to fine-tune the graph and resolve entities, so the model can tell a competitor from a partner and knows the four patterns Palo Alto shows up in.
The outcome was a technical discovery playbook, stage by stage.
Before the demo: current tech stack, network status, security status, prior breaches, the future stack the customer is reaching for, additional server requirements.
At POC: requirements and completion status, budget, product interest, and pilot test cases defined upfront rather than after the pilot starts.
After every call, a score on whether the engineer actually did it.
Chitresh's ask went straight into the same graph. He wanted the system to hear a competitor and check the response against Versa's own battlecard:
This kind of judgment lives in a good engineer's head and not in datasets, which is why it has to be taught rather than mined. The signals were whatever Versa's own leaders said good looked like. Versa was executing its own strategy, not a vendor's best practice.
A playbook nobody runs is not a playbook. Adoption at Versa meant two things: engineers got the right thing at the right moment without opening anything new, and leaders could see it happening.
For the engineers, the behaviors arrive where the work already happens.
The checklist fires off the Salesforce opportunity stage, so technical discovery questions appear on technical discovery calls without anyone switching it on.
Prep notes carry more than what the BDR and AE established. They add the move: the objection-handling script for the competitor already in play, what to concede, what to hold.
Pipeline review surfaces objection flags per deal with the receipt attached: the call, the moment, the customer's own words.
Nothing lives in a new tool. It lands in Teams and in Salesforce.
The shift that mattered most was that people stopped treating it as monitoring.
For the leaders, the problem was the opposite: Chitresh could not use what he had. Sitting at the top of four regions, he saw everything at once in the org-level view:
What he wanted instead was narrow and specific. Zime replaced the dashboard with reports cut to his shape: one roll-up per region, because he coaches those leaders separately; the 20 deals out of 250 worth a fifteen-minute review. They arrive by email on a schedule, so he can read them on a flight with no Wi-Fi.
Right judgment today becomes wrong the moment a competitor ships something. This is where an internal build quietly turns its owner into a maintenance engineer.
The harness keeps the Behavior Graph in production: versioned playbooks that change when the product line or the competitor does, entity resolution so a new competitor name does not break the analysis, and skills for the repetitive work: the AE-to-SE handover, the follow-up email, the RFP response drawn from approved content.
Governance is what makes visibility acceptable. Engineers see their own work, regional leaders see their region, and a small group, Chitresh among them, sees across all of it. That is the difference between inspection and surveillance.
Cost has to be legible too. If every project consumes credits, someone will want to know which ones are expensive and whether costs fall over time. Chitresh, Versa's team noted, is the one who will ask.
Technical wins up 12%, independently validated by HFS Research.
The mechanism is not mysterious. When an engineer walks into a technical evaluation already knowing the customer's stack, the competitor in the room, and the test cases that have to pass, fewer evaluations die on an objection nobody saw coming. Sixty percent of Versa's fatal objections were technical or product-fit, so that is the category these wins came out of.
24% less leakage in the committed pipeline. Deals that survive qualification are now ones an engineer has actually pressure-tested, so less of the committed number turns out to be the fiction Martin described.
At the start, only 20% of SEs were running discovery properly. Within three months, 79% of SEs were doing it, including the upfront POC success criteria that used to get settled late. That is adoption counted as behaviors run on live deals, tied to the stage that authorized them. Not logins, not training completions.
HFS Research named Zime to its 2026 Services-as-Software Hot Tech list. The report, HFS Services-as-Software Hot Tech: Zime AI, published 8 July 2026 by David Cushman, carries the Versa case study.
Your SEs run the calls that decide your largest deals, and their managers cannot attend without it feeling like surveillance.
Most of the objections killing your deals are technical or product-fit, and every rep answers them differently.
You have built agents, a context graph, or a search tool, and the field still does not run what it produces.
Battlecards and playbooks go stale faster than anyone can maintain them by hand.
Zime's Behavior Graph encodes how your best engineers actually run a technical evaluation, elicited from them rather than mined out of transcripts, and delivers it as behaviors in the tools your team already uses.
Book a demo and see it run on your own calls.
