Most companies experimenting with AI in their go-to-market motion are stuck in the same place. The pilot went well, but revenue impact remains unclear. Tools are deployed, experiments look promising, and yet nothing has fundamentally changed about how the team forecasts, executes, or grows.
In this conversation, Justin Shriber (CEO & Co-Founder, Terret) and Andrew Kodner (GTM Business Transformation, DocuSign) dig into why that gap exists — and what it actually takes to cross it.
Revenue leaders are under more pressure than ever. Grow faster, but with fewer resources. Build a board presentation with data that proves the strategy. Turn around a struggling region without starving the ones that are working. Figure out why a big deal slipped at the end of the quarter — and make sure it doesn't happen again.
These aren't AI problems. They're business problems. And as Andy puts it, the mistake most teams make is starting with "let's add AI" instead of starting with the question they need answered.
Even without AI in the picture, most revenue organizations struggle to connect what leadership decides with what reps actually do. Strategy gets built top-down in slide decks that end up in the dust bin. RevOps spends weeks operationalizing new KPIs with data scattered across systems that don't talk to each other. Sellers on the front line can't pass what Andy calls "the Tuesday night test" — can they prep for tomorrow's call start to finish, on their own, without help?
AI hasn't fixed this. In many cases, it's made it worse by flooding teams with more information without giving them more clarity.
Before AI can answer meaningful questions about your business, it needs complete context. For most companies, that context is fragmented across CRM, email, call transcripts, billing systems, and data warehouses.
Justin introduces a concept called the Revenue Graph — a living map that assembles every revenue signal, interaction, and outcome into a single connected view. Think of it as dumping a 2,000-piece puzzle across the room and having AI collect every piece and assemble the full picture.
But building it isn't trivial. Three challenges trip up most teams:
Access controls. Revenue data is locked down differently in every system. Email is user-based. Salesforce is hierarchical. Your data warehouse might use role-based permissions. The Revenue Graph has to respect all of those while still making the right information visible to the right people.
Data association. When a customer sends an email that says "pull the AMIA piece out and we're in," a human knows exactly what that means. A machine doesn't. AI has to infer which deal, which account, and which context that email belongs to — using sender history, timing, textual clues, and past activity patterns.
Context compression. One customer's Revenue Graph totaled 13 terabytes of data — roughly the equivalent of every book in the Library of Congress. An LLM context window holds about 400 kilobytes, or one long email with a couple attachments. Bridging that gap requires intelligent chunking, pre-processing, and summarization before anything reaches the model.
Justin walks through a real customer example. The company had thousands of deals in Europe and watched win rates decline. The RevOps leader asked one question: "Why are win rates dropping in EMEA?"
In 30 seconds, the LLM — plugged into the company's Revenue Graph — came back with a full analysis. It identified 73 lost deals from Q1, surfaced six specific failure points, and backed each one with actual language from real sales conversations. No generic framework imposed on top. No assumptions going in. Just the data, analyzed and explained.
The analysis was only step one. The next question: "What are our closers doing differently?"
The system analyzed the deals that were won, identified the patterns and language that correlated with higher close rates, and generated a playbook. Not a theoretical best-practices doc — a playbook built from what reps actually said in winning conversations.
From there, each rep got a pre-call script customized to their specific deal, buyer, and stage. If Sarah is the decision-maker and Marcus is her CISO, the script anticipates what they're likely to say and maps branch logic for how to respond. Every objection is pre-loaded with counter-language pulled from how your top performers handled similar situations.
After the call, the rep gets a scored evaluation. Here's what you did well. Here's where you missed. Here's how to get from a 7 to a 10.
Justin's advice: don't put this in the rep's ear during the call. Think of it like the NFL — the running back studies the defense before the snap. You don't coach them play-by-play in real time. The preparation is what makes the in-the-moment execution instinctive.
The session closes with a practical framework. Take a 100-rep organization that needs to grow 25%. The traditional playbook is linear: hire 25 more reps, add a RevOps person, add an enablement person. That's $5.6M+ in new headcount costs before you've booked a single dollar.
The alternative: keep the same team and increase their productivity through activated playbooks, better deal execution, and higher close rates. The combined impact is $6.5–8M in cost reduction and revenue gains. Whether the exact number is $7.5M or $3.5M for your business, the directional math is clear — and your competitors are already running it.
Whether you're a CRO trying to figure out where AI fits in your revenue strategy, a RevOps leader evaluating build vs. buy on your data foundation, or an operator wondering how to finally get past the pilot stage — this is a practical, experience-driven conversation worth the hour.