> ## Documentation Index > Fetch the complete guide index at: https://www.synscribe.com/agentic-discovery/llms.txt > Use this file to discover all pages before exploring further. --- title: "Agentic Discovery: The SEO & GEO Playbook for AI Agents (2026)" description: "How AI agents find, evaluate, and pick products. Proven on coding agents (Claude Code, Codex, OpenCode) choosing APIs, MCP servers, and libraries. 12 plays (11 evidence-backed + the new ARD spec), original research, honest verdicts." slug: /agentic-discovery series: The Agentic Discovery Playbook · Hub last_verified: 2026-06-12 --- # The Complete Playbook to Agentic Discovery > **In short:** Agentic discovery is how AI agents find, evaluate, and choose products on behalf of their users. It plays out across the agent's whole journey: what the model already knows, the open web it searches, the docs it pulls in, and the rules installed in its workspace. All of it can be engineered today. This volume proves the playbook on coding agents (Claude Code, Codex, OpenCode) choosing APIs, MCP servers, and libraries. You get the research, an audit you can run in 30 minutes, eleven evidence-backed plays plus the new ARD frontier play, and the myths our data busted. ![Three eras of getting found, side by side. SEO (since ~2000): a human types a query into a search engine; you optimize to rank your pages and earn the click; what you win is POSITION. GEO / AEO (since ~2023): a human asks an AI such as ChatGPT, Perplexity, or AI Overviews; you optimize to be the source the answer cites; fewer clicks, still a reader; what you win is the CITATION. Agentic Discovery (now, unowned): an AI agent searches, evaluates and chooses with no human reading; you optimize to be the default it picks and invokes, end to end; what you win is SELECTION.](/agentic-discovery/images/seo-geo-agentic-discovery.svg "SEO won the click. GEO won the citation. Agentic discovery wins the selection: the product the agent picks without a human in the loop.") Agentic discovery is how AI agents find, evaluate, and choose products, with no human in the loop. This playbook proves it on the tier deciding millions of times a day: coding agents like Claude Code, Codex, and OpenCode choosing APIs, MCP servers, and libraries. The plays are built to travel, and work agents are next. You get twenty-three chapters, twelve plays, and eight tools you can run today, in order, from one B2B studio that ran the playbook on itself. AI agents now read your docs, pick your competitors, and write the integration. All of this happens before a human ever sees your homepage. A two-year-old auth library is the #2 most-fetched documentation source among coding agents, ahead of React. A one-paragraph rules file flipped an agent's product choice 100% of the time in our trials. And the advice flooding your feed about how to win this channel? We pressure-tested it. A lot of it did not survive. This guide is the step-by-step playbook for becoming the product agents choose by default. It is built on original research: live rankings data, an 18-product audit, controlled experiments, three instrumented agent research runs captured with [Birdseye](/agentic-discovery/resources/birdseye), and an evidence-tiered map of every registry that claims to matter. **This guide practices what it preaches.** It ships its own [llms.txt](/agentic-discovery/llms.txt), serves every page as markdown, and stamps every claim with a verification date. We ran the playbook on the playbook. --- ## Section I: Understand **Part 1 · [What Is Agentic Discovery (and Why It's Eating SEO)](/agentic-discovery/what-is-agentic-discovery)** The four-channel model (prior → search/fetch → retrieval → environment), the vocabulary cleanup (GEO vs AEO vs AX), and why 73% of agent doc-traffic comes from terminals, not browsers. **Part 2 · [How AI Agents Actually Choose Products](/agentic-discovery/how-ai-agents-choose-products)** The research: how agents search the open web (a 57× range; ~6% of results opened; ~48% of claims killed), our flip experiments, the live retrieval rankings, and the anomalies. Among them: the two-year-old #2 and the filename that doubles as a growth strategy. ## Section II: Audit **Part 3 · [The Agent-Readiness Audit: Score Yourself in 30 Minutes](/agentic-discovery/agent-readiness-audit)** Eight pass/fail properties, a 0–100 rubric, and exactly what good and failing look like. ## Section III: The Playbook **Part 4 · [The 12 Plays: Overview & Sequencing](/agentic-discovery/agentic-discovery-playbook)** *(start here for the whole method in ten minutes).* An agent runs four steps to pick a product: **find → research → shortlist → act**. Most plays are tools that serve more than one step. Read the rows: | Play | Find | Research | Shortlist | Act | |---|:--:|:--:|:--:|:--:| | [Agent web search](/agentic-discovery/ai-agent-web-search-and-fetch) | ● | ● | | | | [Registries & directories](/agentic-discovery/ai-agent-registries-and-directories) | ● | ● | | | | [MCP server](/agentic-discovery/mcp-server-distribution) | | | ● | | | [Skills & AGENTS.md](/agentic-discovery/agent-skills-and-agents-md) | | | ● | ● | | [llms.txt](/agentic-discovery/llms-txt) | | ● | | | | [Markdown docs](/agentic-discovery/markdown-docs-for-ai-agents) | | ● | ● | ● | | [Snippet density](/agentic-discovery/code-snippets-for-ai-agents) | | ● | ● | | | [Directive layer](/agentic-discovery/stop-ai-using-deprecated-apis) | | | | ● | | [Scaffolder rules / CLAUDE.md](/agentic-discovery/scaffolder-rules-claude-md) | | | | ● | | [Keyless onboarding](/agentic-discovery/agent-first-onboarding) | | | | ● | | [Evals & leaderboards](/agentic-discovery/ai-evals-and-leaderboards) | ● | | | ● | | [Agentic Resource Discovery (ARD)](/agentic-discovery/agentic-resource-discovery) · *frontier* | ● | | | | ## Section IV: Prove **Part 5 · [Measuring Agentic Visibility](/agentic-discovery/measure-ai-visibility)** A metric for each of the four stages, the weekly tracker, how to instrument an agent's own search run, and why citation dashboards miss the layer that matters. **Part 6 · [GEO Myths: What Everyone Says To Do (and What Our Data Shows)](/agentic-discovery/geo-myths-what-doesnt-work)** Fourteen popular claims, four honest verdicts: busted, unproven-tracking, validated-with-conditions, harmful. ## Section V: Case Studies **[Ten Company Teardowns: Overview](/agentic-discovery/case-studies)**: who already did it, with receipts. Three breakouts, three engineered mechanisms, two incumbent paradoxes, two live tensions. - **Breakout · [Better Auth](/agentic-discovery/case-studies/better-auth)**: #2 docs source at two years old, with no training-data prior. - **Breakout · [DodoPayments](/agentic-discovery/case-studies/dodopayments)**: won the "payments" query Stripe never claimed. - **Breakout · [Resend](/agentic-discovery/case-studies/resend)**: the smallest corpus with the highest benchmark score. - **Mechanism · [Bun](/agentic-discovery/case-studies/bun)**: the growth hack that is a filename. - **Mechanism · [Next.js](/agentic-discovery/case-studies/nextjs)**: every play running at once. - **Mechanism · [Convex](/agentic-discovery/case-studies/convex)**: the eval-driven operator. - **Incumbent · [Stripe](/agentic-discovery/case-studies/stripe)**: fighting its own training-data ghost. - **Incumbent · [Tailwind](/agentic-discovery/case-studies/tailwind)**: top-10 with zero agent surface, and why that path is closed to you. - **Tension · [Clerk](/agentic-discovery/case-studies/clerk)**: docs written for agents, not developers. - **Tension · [Drizzle](/agentic-discovery/case-studies/drizzle)**: 440 snippets that beat 2,297. ## Section VI: Reference **[Glossary](/agentic-discovery/glossary)**: 37 standalone definitions. · **[The Data Room](/agentic-discovery/data)**: every number in this guide, dated, with methodology, plus the living Pressure-Test Ledger. **Tracker · [ARD / ai-catalog.json Adoption Tracker](/agentic-discovery/ard-adoption-tracker)**: a live, dated census of who actually ships an ai-catalog.json. Latest snapshot: 0 of 39 top sites, including all eleven spec authors. Per-company results, methodology, and an open reproduction script. **Report · [How AI Agents Actually Search: The 3-Experiment Report](/agentic-discovery/agent-search-experiments)**: the raw evidence behind the search/fetch surface. Three instrumented runs, with every query, fetch, and verdict. --- ## What's published & what's next Every agent that chooses runs the same flow: find → research → shortlist → act. **Volume 1 covers coding agents choosing APIs, MCP servers, and libraries. It is published and tested:** three instrumented research runs, controlled flip experiments, live retrieval rankings, and every claim dated in the [Data Room](/agentic-discovery/data). The next volumes bring the same research depth to the rest of the territory. **Volume 2 instruments work agents choosing SaaS connectors**, with shopping and procurement agents behind it. Most of the levers carry over: registries, llms.txt, markdown docs, MCP, and evals serve any agent tier. The twelfth and newest play is the one frontier in the set: **[Play 12 · Agentic Resource Discovery (ARD)](/agentic-discovery/agentic-resource-discovery)**, Google's new `ai-catalog.json` spec, which could reshape the *find* step. We've published the complete field-by-field guide and our own day-one census, and we separate honestly what's true today (a cheap file to ship) from what's still a bet (a registry an agent actually queries). As of 2026-06-18, 0 of 39 top sites, including all eleven companies that wrote the spec, publish one. It's sequenced last on purpose: run the eleven evidence-backed plays first. ![The agentic-discovery territory: the find → research → shortlist → act flow mapped against four tiers of agents (coding agents such as Claude Code, Codex, OpenCode, plus work agents, consumer agents, and vertical enterprise agents). The coding-agent column is shaded published and tested; the remaining tiers are next on the publishing roadmap.](/agentic-discovery/images/diagram-discovery-matrix-hub.svg "What's published (coding agents, Vol. 01) and what's next (work agents, Vol. 02, same research depth).") --- ## Free agentic-discovery tools you can run today Open, no email gate. **Mac App · [Birdseye](/agentic-discovery/resources/birdseye)**: Birdseye is our free agent-observability app for the Mac. It replays a coding agent's research run as four debuggable layers: every search it typed, every page it opened (or only saw), and every claim it kept or killed. It is the instrument behind every number in this guide: the 57× search spread, the ~6% fetch rate, the ~48% claim-kill rate. Point it at your own category to see whether agents pick you. - **Worksheet · [Agent-Readiness Scorecard](/agentic-discovery/resources/agent-readiness-scorecard)**: Run the Part 3 audit yourself. Eight checks, one score. - **Template Pack · [llms.txt Template Pack](/agentic-discovery/resources/llms-txt-template-pack)**: Annotated and minimal llms.txt, five directive snippets, a linter. - **Runbook · [Registry Submission Runbook](/agentic-discovery/resources/registry-submission-runbook)**: Four Tier-1 checklists and a tracking grid. - **Prompt Pack · [Deprecation Eval Prompt Pack](/agentic-discovery/resources/deprecation-eval-prompt-pack)**: Twenty prompts to catch models shipping your old API. - **Templates · [Rules-File Templates](/agentic-discovery/resources/rules-file-templates)**: CLAUDE.md and AGENTS.md starters, ethics box included. - **Repo · [Eval-Harness Starter](/agentic-discovery/resources/eval-harness-starter)**: A skeleton to score yourself the way leaderboards do. - **Spreadsheet · [Visibility Tracker Template](/agentic-discovery/resources/visibility-tracker-template)**: 66 formulas, zero setup. Track share-of-model weekly. --- ## Frequently asked questions **What is agentic discovery?** Agentic discovery is how AI agents find, evaluate, and choose products on behalf of their users. It is the full flow from first awareness to completed integration or purchase, with no human in the loop. It runs across four channels: the model's training prior, the web search-and-fetch surface, the retrieval layer (llms.txt, Context7, MCP), and the environment layer (AGENTS.md rules files). **How is agentic discovery different from GEO or AEO?** GEO and AEO optimize the answer an AI shows a human. That is a citation layer. Agentic discovery covers what happens when the agent itself decides and acts: the queries it writes for itself, the pages it opens, the claims it verifies, and the product it silently installs. Mention is not selection, and this guide measures selection. **Which agents does this playbook cover today?** This edition is proven on coding agents (Claude Code, Codex, OpenCode, Cursor, Copilot) choosing APIs, MCP servers, libraries, and developer tools. Roughly 60–70% of the levers (registries, llms.txt, markdown docs, MCP, evals) serve any agent tier. Volume 2 extends the instrumentation to work and computer-use agents. **What is Birdseye?** Birdseye is our free Mac app for agent observability. It replays an AI coding agent's research run as four layers (every query, every fetch, every extracted claim, every verdict) so you can see exactly why an agent did or did not pick you. It produced every search/fetch statistic in this guide. **What is Agentic Resource Discovery (ARD) and should I publish an ai-catalog.json?** ARD is an open spec from Google and a Linux Foundation working group (announced late May 2026) for discovering agentic resources: you publish a `/.well-known/ai-catalog.json` listing your MCP servers, agents, and APIs, and registries index it so agents can find you. If you ship one of those resources, publishing the file is a cheap (~1 hour) first-mover option, our 2026-06-18 census found 0 of 39 top sites doing it yet. But there is no agent-queried ARD registry today, so treat it as a forecast, not a working channel. Full guide: [Agentic Resource Discovery (ARD)](/agentic-discovery/agentic-resource-discovery). --- ## Get a diagnosis **Are you the default an AI agent reaches for?** Get an agent-readiness diagnosis of your product (especially API and MCP products) and the actionable punch-list to become the one agents pick by default. [Get a free diagnosis →](/contact) · or [subscribe to research](/playbook) for what we publish next. --- ## The research behind this guide 18 products audited at the docs/registry/MCP layer · live retrieval rankings pulled and trended · controlled agent experiments (selection flips, stale-window tests) · three instrumented agent research runs captured with [Birdseye](/agentic-discovery/resources/birdseye), tracing every query, fetch, and verdict · every agent registry tiered by measured impact, not server counts · all claims dated, all limitations stated. Numbers live in the [Data Room](/agentic-discovery/data). Methodology and biases are published in full, so quote us, check us, or replicate us. --- *Last verified 2026-06-12. We re-test the claims in this guide quarterly, and changes are logged in the [Data Room](/agentic-discovery/data).* Next: [Part 1: What Is Agentic Discovery →](/agentic-discovery/what-is-agentic-discovery) > **Stay ahead of the agents.** We re-test this playbook quarterly and publish what changed: new data, busted myths, ranking shifts. [Get the update digest →](/agentic-discovery#updates) > > **Want this done for you?** Synscribe runs agentic-discovery programs for B2B SaaS and developer platforms. [Talk to us →](/contact)