xiry.

Marketing & Sales · Framework

Agent-led Growth Playbook

A framework for optimizing discovery and onboarding for autonomous AI agents.

From the Xiry library · Updated 11 October 2026

A go-to-market playbook from Browserbase detailing how to drive signups and revenue by treating AI agents as primary users. It divides the strategy into Agent Engine Optimization (AEO) for discoverability and Agent Experience (AX) to minimize integration friction.

The framework

Results in 3 months at Browserbase: doubled customer base, +119% self-serve revenue, +10% activation. ALG = AEO (discoverability, drives signups) + AX (agent experience, drives activation).

CONTEXT: 37% of consumers start search with an AI tool (~50% Gen Z); ~60% of Google searches end without a click; LLM-referred visitors convert at 1.66% vs 0.15% from traditional search (~10x). Hard to track: same LLM question gives same answer only ~73% of the time; existing AEO tools fire thousands of prompts and count mentions. Agents lean on training data first, web search only when short — so AEO is basically good SEO.

AEO — 3 things that moved the needle:

1. Be everywhere. Only way into model weights is presence across the public web before the next training run; recommendations track frequency. They did 26 launches in Q1 (+64% QoQ).

2. Programmatic use-case pages. Win 'open' searches where the prompter knows only the outcome ('scrape this site for me'), not the product category. Built one page per use case an agent might seek. An internal agent skill spins up a new page daily, approved in Slack; the skill was given brand guidelines, platform story, blogs for style, and a Semrush API key for its own research. /use-case/agents tripled traffic in a month; best performers now run as ads.

3. Reddit. Most-cited domain by AI models (Semrush). Started Q1 with zero presence; one month, 4 threads: 8,700 views, 46 upvotes, 117 comments, 3 of 4 rank 1–3 on Google.

AX — 'Instant Churn': switching used to cost a rewrite, now it costs a sentence; agents discard tools whose setup fights them. Whole game is time-to-aha.

- Treat the agent as a user and measure it. Made Claude the most important customer; worked with 2027.dev running agent harnesses (Claude Code) in a sandbox against the docs; built own evals (Skill Test Arena, batch runs, DX-Eval vs Skill Test modes).

- Fixed for agents: /quickstart and /quickstart/node 404'd (agent bounced through 5 pages); package name hallucinated as browserbase-sdk / browserbase-js — registered both wrong names, pointed them at the real @browserbasehq/sdk.

- Right already: llms.txt, Context7 indexing, official MCP server, full TypeScript types, published OpenAPI spec (5/5 discoverability). 88/100 on Mintlify's agent benchmark; won 2027.dev's arena; ~90 seconds from 'what is Browserbase' to working integration.

- Homepage changes: added a skill.md route (ready-made agent prompt), agent-first onboarding flow, main CTA changed from 'Contact sales' to 'setup for agents' — climbed into top 5 most-clicked homepage buttons within a week.

Closing thesis: the agent is the new top of the funnel; every GTM team will build for it.

Sources cited: AdaSearch (arxiv.org/pdf/2512.16883), gen-optima.com LLMR, semrush.com/blog/most-cited-domains-ai, ziptie.dev AI-vs-Google, searchengineland zero-click, demandlocal.com AI conversion benchmarks.

Video walkthrough (404s): shows the Browserbase homepage ('Give your agents access to the whole web'), the healthcare compliance page, the Webflow CMS of ~20 /use-case/ pages (AI Agents for Sales/Marketing/Ecommerce/Real Estate, Health, Compliance…), the internal Skill Test Arena eval UI (Repeat Count, Batch Name, Evaluation Mode, Model, Auto-Improve, Start 5 Runs), and their internal dashboards: Social buzz chart (huge spike from an @levelsio tweet), Web traffic, Top referrers, Signups, Activation & engagement — everything pulled through an API to track and report numbers.

What it's for

  • Optimize technical documentation and SDK naming to prevent agent integration failures
  • Scale programmatic landing pages targeted at outcome-focused LLM queries
  • Evaluate developer documentation friction using automated agent testing harnesses

Key points

  • AEO strategy focusing on broad web presence and Reddit discussions to influence model training data
  • Programmatic CMS use-case page generation targeting intent-based queries
  • Agent onboarding optimizations including llms.txt, MCP servers, and skill.md endpoints
  • Defensive package naming aliasing commonly hallucinated SDK names to real packages
  • Agent evaluation harness to test and benchmark time-to-integration for AI agents

Pricing

Free

Details

Type
Framework
Category
Marketing & Sales
Topic
Growth Frameworks
Open source
No
Platforms
Web
For
Growth engineers, technical marketers, and developer relations teams
Tags
growth, aeo, agent experience, developer marketing, ai agents

Why it was saved

  • @JaySahnan, 20 July 2026: Concrete, metric-backed playbook for a brand-new GTM channel with reusable tactics and numbers.

Related

Updated