For decades, go-to-market has followed a familiar pattern: attract leads, qualify them, move them through stages, close deals. It is linear, structured, and deeply embedded in how companies operate. And it is breaking — not because it stopped working, but because something fundamentally better is emerging.
Just as Harvey redefined legal workflows by placing AI at the centre of how legal work actually gets done, GTM is now undergoing the same transformation. The result is a shift from the traditional funnel to what can only be described as an Unfunnel — an AI-native, always-on, continuously adapting revenue engine. This briefing maps the new five-layer architecture, contrasts it with the funnel it replaces, sets out the new metrics that go with it, and offers a 90-day pragmatic way in.
The funnel was built for a different era
The classic GTM model looks logical from the outside — a clean, sequential pipeline that turns traffic into revenue:
Underneath, it is riddled with inefficiencies. Fragmentation — marketing, sales, and operations run in silos with separate tools and dashboards. Manual handoffs — leads pass between teams, losing context at every transition. Batch processing — campaigns and outreach happen in waves, not continuously. Lagging signals — by the time a human interprets the data and acts, the moment has passed. Stage fiction — prospects rarely move in the clean, sequential way our CRMs assume.
Most importantly, the system is human-orchestrated. Every step depends on people navigating tools, interpreting data, and stitching workflows together manually. That made sense when software was the bottleneck. Today, it is not.
The shift — AI as the operator, not the assistant
The real transformation in GTM is not about adding AI features to existing tools. It is about changing who, or what, runs the system.
Instead of humans coordinating across a dozen point solutions, AI becomes the central operator — ingesting signals, making decisions, and executing actions across the stack. This is the same shift Harvey introduced in legal: not replacing systems of record, but becoming the layer where work actually happens. The CRM does not disappear; it stops being the interface. Users do not live inside the CRM anymore. The AI does.
The Unfunnel — a new five-layer GTM architecture
The Unfunnel replaces linear stages with a continuous, signal-driven system — five layers, one direction of flow, one feedback loop.
Funnel vs. Unfunnel — a side-by-side
Nine dimensions, two architectures, one direction of travel. Read top to bottom for the structural argument.
| Dimension | Traditional Funnel | The Unfunnel |
|---|---|---|
| Starting point | Leads | Signals |
| Cadence | Batch / campaign cycles | Continuous · always-on |
| Orchestration | Humans across tools | AI as central operator |
| Progression | Fixed stages (MQL → SQL → Opp) | Fluid · signal-driven |
| Personalisation | Mail-merge tokens | Context-aware, per account |
| Interface | CRM dashboards | AI layer · CRM as memory |
| Key metric | Lead volume & conversion rate | Account progression & signal coverage |
| Team shape | Large SDR bench, siloed teams | Lean, judgement-heavy, cross-functional |
| Optimisation | Quarterly reviews | Real-time · self-correcting |
The shift is not incremental. It is structural — touching the data model, the orchestration logic, the team shape, and the metrics that get reported to the board.
The end of funnel stages. MQL, SQL, and Opportunity are replaced by continuous account progression. Accounts move fluidly based on real-time signals, not arbitrary stage definitions a RevOps lead drew up in 2019. Always-on GTM. Quarterly campaigns become obsolete. Companies operate in a state of continuous engagement. The system never stops running. True personalisation at scale. Not mail-merge with a first name — deep account understanding, context-aware messaging, dynamic content, at a volume no human team could match. Human leverage increases. Roles do not disappear; they evolve. SDRs get redefined or consolidated. AEs focus on closing and strategic relationships. Marketing becomes a signal and content engine. AI handles execution; humans focus on judgement, creativity, and trust.
New metrics for a new model
If the funnel is dying, so are the metrics built around it. Volume KPIs — MQLs per month, dials per day — were proxies for human effort. In the Unfunnel, what matters is different.
Signal coverage — what percentage of your TAM are you receiving usable signals from? Time-to-action — from signal detected to first relevant touch. Account progression velocity — how quickly accounts move toward buying readiness. Revenue per AI action — efficiency of the operator layer, not the humans. Message-to-context fit — relevance scoring of outbound touches. Feedback loop latency — how fast learning from won and lost deals reaches the operator layer.
If your dashboard still leads with "MQLs generated this month," you are measuring the wrong thing. The metrics that tell you whether the Unfunnel is working are not the metrics most boards are reviewing today.
Why most companies will struggle
The biggest mistake companies will make is trying to bolt AI onto existing workflows, keep the CRM at the centre, and preserve funnel thinking. That path leads to incremental improvements — not transformation. The Unfunnel requires re-architecting GTM with AI as the control layer. That is an organisational change, not a tooling change.
Expect friction around five fronts. Data readiness — fragmented, dirty data starves the operator layer. Incentive design — comp plans still reward lead volume and activity, not signal-driven outcomes. Org structure — SDR/AE/Marketing lines blur when AI handles execution. Change management — sellers who grew up in the funnel will resist losing control of their book. Risk and trust — brand and compliance rightfully ask: what is the AI actually sending?
Companies that treat this as a RevOps project will fail. Companies that treat it as a GTM operating model redesign will win.
A 90-day pragmatic starting point
You do not need to boil the ocean. A grounded way in — one motion, narrow scope, fast feedback.
- Pick one motion. Outbound to a defined ICP, expansion within existing accounts, or churn prevention. One.
- Inventory your signals. List everything you already collect. Most teams are surprised — you have more than you think, you just are not using it.
- Define three to five trigger-to-action pairs. For example: "New VP of Engineering arriving + product usage drop → CSM alert + tailored executive email." Specific. Testable.
- Put a human in the loop — at first. Review every AI-generated action for the first few weeks. Builds trust. Surfaces where the AI is wrong before it scales.
- Measure against the new metrics. Time-to-action, reply rates, progression velocity. Not lead volume.
- Expand the loop. Once one motion is humming, extend to the next. Compounding beats moonshots.
The goal is not a moonshot replatform. It is proving the Unfunnel works on a narrow slice — and then compounding.
A simple mental model. Most companies today are busy upgrading their memory. The winners will own the brain.
The operator's takeaway
Harvey did not replace legal systems. It changed how legal work gets done. The same thing is now happening in GTM. The funnel is not disappearing because it was wrong — it is disappearing because it is no longer necessary. What replaces it is something more powerful: a continuous, intelligent, AI-driven system that does not move prospects through stages.
That is the Unfunnel.
If you are rethinking your GTM architecture, the question is not whether to adopt AI. It is whether you will treat it as a feature — or as the operator.