DISCOVER+ SEARCH
Core questionIs this for me?
What the agent doesResearches and shortlists options
- Key touchpoints
- Website
- Social
- Events & conferences
- Community (e.g., Discord, Slack, Reddit)
- AI answers
- Registries
MilestoneInterested
A developer journey map for the AI agent era: how developers, vibe coders, and other builders, with their AI agents, go from first search to scale.
Release 2.0 · Oct 2026 · See what's new
Core questionIs this for me?
What the agent doesResearches and shortlists options
MilestoneInterested
Core questionCan I trust it?
What the agent doesRuns a trial. You approve the fit.
MilestoneSigned up
Core questionCan I get started?
What the agent doesReads the docs and sets up
MilestoneReady to build
Core questionDoes it work?
What the agent doesWrites and tests code. You review.
Milestone★ Hello World → First success
Core questionCan we grow with it?
What the agent doesHandles routine tasks. You own incidents.
MilestoneIn production
Click any stage to see why it matters and a quick check for your team.
↺ It's not a straight path to adoption: builders who hit friction loop backwards or walk away, and happy users become your advocates.
Is this for me?
Most builders now ask an AI assistant before they ever visit your site. If it doesn't know you, or describes you wrong, you never make the shortlist.
Researches and shortlists options
Nobody knows you yet. Their AI tool doesn't mention you, or gets you wrong.
Ask three AI assistants what tool to use for your use case. Are you named, and described correctly?
A plain-language guide to each stage of the Developer + Agent Adoption Journey.
Why it matters. Most builders now ask an AI assistant before they ever visit your site. If it doesn't know you, or describes you wrong, you never make the shortlist.
Where builders get stuck. Nobody knows you yet. Their AI tool doesn't mention you, or gets you wrong.
Key touchpoints. Website, social, events and conferences, community (Discord, Slack, Reddit), AI answers, registries.
Quick check for your team. Ask three AI assistants what tool to use for your use case. Are you named, and described correctly?
Why it matters. Developers want to try before they buy, and agents need to read your limits and pricing. Any gate here, like a sales call or a hidden price, sends them to a competitor.
Where builders get stuck. Can't register, no free way to try it, unclear cost to level up.
Key touchpoints. Tiered pricing (free to paid), community, sign-up, sandbox, reviews, benchmarks.
Quick check for your team. Can someone sign up, see what they get free, and know what it costs to level up, in under five minutes?
Why it matters. Docs now have two readers: the person and their AI agent. If the agent can't read your docs, it writes code that doesn't work, and the person blames your product.
Where builders get stuck. Slow to get a key, docs agents can't read, no clear quickstart.
Key touchpoints. Quickstart, docs (llms.txt, markdown), tutorials, docs AI assistant, skill files.
Quick check for your team. Point an AI coding tool at your quickstart. Does it reach Hello World without help?
Why it matters. This is where Hello World turns into a first real success. Clear errors, current examples, and tools agents can call decide whether the build finishes or stalls.
Where builders get stuck. Unclear errors, outdated code samples, first run fails.
Key touchpoints. Coding tools (Cursor, Claude Code, Copilot), SDKs, MCP tools, error messages, changelog.
Quick check for your team. When a first run fails, does the error tell a person and an agent exactly how to fix it?
Why it matters. Adoption is won or lost after launch. Predictable bills, responsive support, and a place for users to show their work turn customers into the advocates who bring the next ones.
Where builders get stuck. Surprise bills, slow support, feedback goes nowhere.
Key touchpoints. Support, status page, showcase, champions, usage dashboard.
Quick check for your team. Do your happiest users have an easy way to share what they built and be recognized for it?
This map is versioned like software. Each release reflects how developers and their AI agents actually adopt tools, and what's changed since the last one.