ESSAY··3 min read
The Execution Layer.
The modern GTM stack makes everything easier except the one thing that matters: deciding what to do next. That's the gap that's finally buildable.
The modern GTM stack makes everything easier except the one thing that matters: deciding what to do next. That's the gap that's finally buildable.
Every decade or so, the revenue operations toolkit gets a new layer. CRM. Marketing automation. CDPs. Revenue intelligence. Reverse ETL. Each wave promised that centralizing the data would unlock a faster, smarter go-to-market motion. And each wave, in its own way, delivered — but the ceiling was always the same. The data got cleaner, the reports got prettier, and the team still spent most of its week doing the operational glue work that makes the reports worth running.
The next layer isn't another tool. It's an operating surface — a conversational layer that sits across the entire GTM stack, reads any signal, routes it to the right system, and acts.
The next generation of revenue operations runs on AI-native infrastructure. Systems that don't just report — they surface, decide, and act.
A rep notices something on a call. A CSM spots churn risk. A marketer sees a segment spike. In theory, those observations flow into the system of record, route to the right owner, and become action. In practice, they evaporate into a Slack message that dies on page 12 of the channel, or a CRM field nobody queries, or a spreadsheet in someone's Downloads folder. Every org I've worked in has a tax that looks like this — and every org assumes it's a tooling problem. It isn't. The tools are fine. What's missing is the layer between the signal and the action.
I've watched strong reps solve this by memorizing which fields matter, keeping shadow spreadsheets, and spending a fifth of their week on operational glue the system should be doing for them. That glue work is the opportunity. It's the loop that hasn't been closed yet.
In the last 18 months, the infrastructure to close those loops went from aspirational to buildable. Large language models can read unstructured signal — calls, emails, notes, docs — and reason over it with a schema. MCP makes the plumbing between systems cheap. Claude Code lets a one-person ops team ship a production workflow in an afternoon. None of this needs a platform team, a budget line, or a vendor RFP. It needs clarity about which loops matter and the discipline to close them one at a time.
That's what everything in this portfolio is. Not a platform, and not a bet on AI in the abstract — a collection of loops closed deliberately. The orchestration layer that dispatches my daily work to specialists. The deck generator that turns three hours of CSM prep into ninety seconds. The attribution pipeline that learned from 50% to 95% accuracy by capturing every correction. The sales intelligence hub that ties the weekly digest to the pre-call brief. Each piece replaces work I used to do by hand.
The interesting question isn't whether AI will change RevOps. It's which loops compound the most when you close them. My shortlist, in order: attribution + data hygiene, pre-call intelligence, post-call signal capture, renewal risk, forecasting. Each one makes the next one easier. Each one moves judgment from the system of record back to the humans who belong there — the sellers, the CSMs, the marketers, the leaders — and lets them operate on leverage instead of entropy.
Teams that treat this seriously will run leaner, ship faster, and have a more accurate picture of their book. The work that was operational tax becomes institutional memory instead — captured once, applied everywhere. That's the shift I'm betting on. It's also the shift that separates the ops leaders who stay in the requester role from the ones who step into the builder role.
If you're building in this direction — or you want to compare notes on where you've landed — find me.