Agents Repeatedly Picking the Same Vendor Is Now a Market-Adoption Signal, Not Just a Feature Win

In his weekly SaaStr recap on this week’s 20VC podcast, Jason Lemkin rounds up a batch of deal and market data and draws a sharper conclusion: when AI agents repeatedly choose the same third-party tool — his example is Clay, now valued at $7B — that repeated selection is itself a market-adoption signal, not just a product win. He points to Nvidia’s 70% growth guidance as the benchmark other AI infrastructure vendors now get measured against, alongside named valuations: Hugging Face at $12.9B, Cognition at $46B (running $800-900M ARR, targeting $1.6B), and Owner at a $2.3B raise.

The build-speed data point is the one worth sitting with: an agentic build on Replit generated 448 open tasks, with feature-shipping cycles compressed from roughly a quarter to about a week — a ~100x speedup. Lemkin also flags a headcount signal: companies growing 100%+ year-over-year increased headcount by 133%, while companies growing under 50% added none. His read on Salesforce specifically is that it’s increasingly consumed “headless” via agents, eroding UI as a competitive moat — “assume any agent with write access will eventually take an action you didn’t ask for,” and “third in the right market pays better than first in a small one.”

The headcount split tracks a pattern Lemkin has been building for weeks. In an earlier SaaStr analysis, he showed the public-SaaS growth distribution compressing hard — only seven public companies growing above 30%, a band now commanding 8x-20x+ revenue multiples against roughly 1.9x for sub-10% growers. And in a follow-up piece, he argued most CEOs claiming to “rebuild for AI” are stuck at roughly 40% of the way there — shipping AI features on top of legacy product, pricing, and sales motion rather than rebuilding any of it.

Read together, the throughline for GTM leaders is that agent behavior and headcount allocation are becoming the leading indicators worth tracking — not because they’re novel metrics, but because they’re catching the compounding-versus-legacy split before it shows up in a quarterly growth number.