I've spent 30 years in enterprise tech - much of it in the seat where the quarter is everything - from PistonCloud through its acquisition by Cisco, MuleSoft through its IPO, Kong from zero to unicorn scale, and Venafi into a new category. You learn what compounds and what does not.

‍
The pattern I kept seeing was that the most important GTM decisions are made early, before they look like GTM decisions. Category frame, first customer, first sales motion, first GTM hire - these choices quietly shape the trajectory of the company.

‍
What pulled me to Z47 was the chance to be useful at exactly that moment. The US-India corridor is where some of the most consequential enterprise AI companies of this decade will be built. I'd rather be in several of those rooms early than one of them late.

The biggest change is that software is no longer the product. Outcomes are. SaaS sold access to tools. AI-native companies sell completed work. That changes everything about GTM.

‍

The old funnel assumptions break down. Buyers now evaluate speed-to-value, autonomy, trust, and operational impact. User and buyer are diverging: AI adoption starts bottom-up while governance and budget move top-down.

‍

Monetization breaks too. SaaS ran on 80% gross margins and priced for value capture. AI runs on 30-60% margins and has to price for variable cost. Per-seat pricing in an AI-native company is either subsidizing usage or capping value - usually both. Credits, usage tiers, and outcome-based pricing are the new default .

‍

What does not change is that great companies still need sharp positioning, emotional storytelling, deep customer understanding, and relentless execution. AI changes the mechanics of GTM but it does not remove the need for conviction, narrative, or trust. If anything, those matter more, now that markets are noisier and buyers are overwhelmed.

The whole idea is wrong, and always was, masked by the ZIRP era. It conflates marketing with demand gen. Demand gen comes after PMF. Positioning, category framing, and the language for the problem come before, they shape which customers you talk to and what signal you read.

‍

I've seen founders declare PMF off twenty design partners who loved the product for twenty different reasons. That's not PMF. That's twenty pilots in search of a category.

‍

Before you scale anything: write the one-page narrative, enemy, misdiagnosis, wedge, proof. Test it on ten target buyers. If they can't repeat your enemy back, you don't have a category yet. Fix it before you hire a single AE.

Three broad categories will define the next decade:

1. AI operational layers: the infrastructure, governance, security, orchestration, observability, and identity stack needed to operationalize AI safely inside enterprises.

2. AI-native workflow and workforce companies: products that don't add AI to existing SaaS but fundamentally collapse labor, coordination, and operational complexity. The winners won't look like SaaS with AI features; they'll look like software replacing the work itself.

3. AI infrastructure beyond the frontier labs: compute, inference, runtime, and silicon-access layers that determine whether enterprise AI scales economically. Most value gets captured outside the frontier model companies.
‍

The biggest shift: defensibility in AI won't come from the model. It comes from distribution, trust, workflow ownership, proprietary context, and speed of iteration. Deep work, not quick fixes.

At 2 a.m. the founder doesn't need a framework, they need to find their "sach" - their own truth. They need someone who'll pick up, listen for ten minutes without interrupting, and then ask two questions: what's the customer actually saying, and what would you do if you had no board, no investors, no ego in the decision?

‍

Founders First means I'm not optimizing for how the call makes me look at the next board meeting. It means I'll tell them when their instinct is right and the new entrant is noise, and I'll tell them when they've been in denial for a quarter. Repositioning is a serious move - get it wrong and you reset the clock. Get it right and you skip a chapter.