Pulse: where are you on AI adoption?
Across the founders and operators we talk to, AI adoption breaks down into four pretty consistent stages. Quick read, quick vote, and a note on the trap that sits inside each stage.
The four stages
These are not maturity-model categories. They are descriptions of what an organization is actually doing on a given Tuesday.
- Exploring. ChatGPT subscriptions floating around. A few experiments in slide decks. No production code calls an LLM yet. Most companies sit here longer than they admit.
- Piloting. One or two narrow workflows are wired up. Maybe a support draft assistant, maybe an internal Q&A bot. They work, mostly. Nobody has seen the real bill yet because traffic is small.
- Scaling. Multiple workflows live. The team has a vendor preference. Someone is starting to think about evals and prompt caching. Costs are real. So is the dependency.
- AI-by-default. New features are shipped with an LLM in the loop without a meeting. The team is not asking "should we use AI here." They are asking "which model and which guardrails."
Cast your vote
No data goes to a server. The tally lives in your browser only. Useful as a mirror, not as a survey.
The trap at each stage
Every stage has a failure mode that comes from the stage itself, not from the technology.
- Exploring. Endless deck cycles. Demos that never become deployments. The trap is mistaking exploration for progress. Cure: pick one workflow with a real metric and ship a thin version.
- Piloting. The pilot looks good and nobody asks how it scales. Then it goes wide and the bill quadruples or quality collapses on edge cases. Cure: evals and a load test before scaling, every time.
- Scaling. Three teams build their own LLM gateway, three different prompt templates, three different vendor accounts. The platform-debt trap. Cure: name an owner and consolidate before you have ten of everything.
- AI-by-default. Speed beats discipline. Things ship without guardrails, redaction, or monitoring because nobody has the headspace. Cure: a small platform team whose job is to make the safe path the easy path.
Curious where the line between piloting and scaling actually sits for you? Take the AI Readiness Quiz. Eight minutes, no signup.