# Lean Startup

> Build-Measure-Learn methodology — an AI agent skill based on *The Lean Startup* by Eric Ries.

Published 2026-02-19 · Updated 2026-08-17 · Product & Innovation · Free and MIT-licensed
Canonical: https://skills.wondel.ai/skills/lean-startup/
Source: https://github.com/wondelai/skills/tree/main/lean-startup

**Install:** `npx skills add wondelai/skills/lean-startup --global`

## What is the Lean Startup skill?

Lean Startup is a free AI agent skill built on Eric Ries’s book. It puts Build-Measure-Learn in your agent’s hands: name the leap-of-faith assumption, design the smallest experiment that could disprove it, choose one actionable metric over the flattering ones, and write down the pivot-or-persevere threshold before the experiment is allowed to run.

## Key concepts from The Lean Startup your agent applies

- **Build-Measure-Learn** — The core feedback loop — build the smallest thing, measure what matters, learn whether to pivot or persevere.
- **Minimum Viable Product** — Ship the simplest version that lets you start learning — a landing page, concierge service, or wizard-of-oz test.
- **Validated Learning** — Track actionable metrics that prove progress toward a sustainable business, not vanity metrics.
- **Innovation Accounting** — Measure progress with leap-of-faith assumptions, cohort analysis, and pivot-or-persevere decisions.
- **The Pivot** — A structured course correction — change strategy without changing vision when metrics show the current path won't work.

## When to use Lean Startup — and when not to

Reach for it when:

- You have a twelve-month build plan and no evidence at all that anyone wants the first month of it.
- Signups rise every week, the chart looks wonderful, and nobody can say whether the business is working.
- The team has argued about a feature for six weeks when a landing page and two hundred visitors would settle it.
- You suspect the strategy is wrong, but “pivot” has become a word nobody will say out loud without a number behind it.

Reach for something else when:

- The experiment wants to be a facilitated week with a prototype and five users in the room — Design Sprint has the schedule.
- The learning has to come from conversations rather than instrumented behavior — The Mom Test is the sharper instrument.
- You already have fit and need a delivery cadence with capped scope — The 37signals Way bets in six-week cycles.
- You need the metric definitions, stage gates and business-model archetypes themselves — that is Lean Analytics.

## About Eric Ries

Pioneer of the Lean Startup movement. Eric Ries is an entrepreneur and author who pioneered the Lean Startup movement. He co-founded IMVU and has advised startups, large enterprises, and venture capital firms on applying lean principles to innovation.

More: https://en.wikipedia.org/wiki/Eric_Ries · https://www.wikidata.org/wiki/Q1752858
The book: https://openlibrary.org/books/OL24982481M/The_Lean_Startup · https://en.wikipedia.org/wiki/The_Lean_Startup · https://www.wikidata.org/wiki/Q7746545

## Example prompts

- Define an MVP scope for my product idea using lean-startup skill _(Product planning)_
- Design a validation experiment for our new feature hypothesis using lean-startup skill _(Experimentation)_
- Set up innovation accounting metrics for our startup using lean-startup skill _(Metrics)_
- Help me decide whether to pivot or persevere based on these metrics using lean-startup skill _(Strategy)_

## Frequently asked questions

### How do I install the Lean Startup skill?

Run `npx skills add wondelai/skills/lean-startup --global`. It takes about 30 seconds and needs no account. The Lean Startup skill then works in Claude, Claude Code, Claude Cowork, Codex, Cursor, OpenClaw and Hermes Agent — anything that reads the open agentskills.io format — and your agent loads it on its own when a task calls for it. It is free and MIT-licensed, and the source is at https://github.com/wondelai/skills.

### Which book is the Lean Startup skill based on?

It packages *The Lean Startup* by Eric Ries — pioneer of the Lean Startup movement. The skill distils the book's method into instructions your agent follows while it works, covering build-measure-learn, minimum viable product and validated learning. It sits in the Product & Innovation part of the library.

### How does this differ from the Design Sprint skill?

A Design Sprint is one event with a fixed five-day agenda; Build-Measure-Learn is a loop you never stop running. The sprint is superb at the first question — does this concept survive contact with five real people — and then it is over. Lean Startup keeps going after launch, on real usage, with cohorts and innovation accounting deciding whether to pivot. Plenty of teams use a sprint to buy the first answer cheaply and this skill to keep earning the rest.

### What actually counts as an MVP?

The smallest thing that produces the learning, which is usually not software. The skill will push you toward a smoke test (a page, a price, a signup button, no product), a concierge MVP (you do the work by hand while the customer experiences a service), or a Wizard of Oz (a real interface with a human behind the curtain). The measure is not how small it is — it is whether a specific result would change what you do next. If nothing would, you are building, not learning.

### Is the book still worth reading?

The skill carries the machinery: the loop, the MVP types, innovation accounting, the taxonomy of pivots from zoom-in to engine-of-growth, and the three engines — sticky, viral, paid. The book is where the argument lives: that a startup is an institution for managing extreme uncertainty, and that most of what feels like progress inside one is not. That framing is the bit people carry for a decade; the mechanics are the bit an agent can apply on a Tuesday afternoon.

### Doesn’t “MVP” just become an excuse for shipping something bad?

Constantly, and no skill can stop you. Ries’s MVP earns its embarrassment only if you wrote down beforehand what result would change your mind. Shipping a thin slice of your full plan, measuring nothing, and calling it an MVP is exactly the behavior the book argues against — you pay the reputational cost and collect none of the information. If your agent cannot state the hypothesis and the threshold, you do not have an MVP, you have a rushed release.

## Related

- All Product & Innovation skills: https://skills.wondel.ai/skills/product/
- Jobs to Be Done: https://skills.wondel.ai/skills/jobs-to-be-done/
- Hooked UX: https://skills.wondel.ai/skills/hooked-ux/
- Improve Retention: https://skills.wondel.ai/skills/improve-retention/
- Design Sprint: https://skills.wondel.ai/skills/design-sprint/
- Continuous Discovery: https://skills.wondel.ai/skills/continuous-discovery/
- Inspired: https://skills.wondel.ai/skills/inspired-product/
- The Mom Test: https://skills.wondel.ai/skills/mom-test/
- The 37signals Way: https://skills.wondel.ai/skills/37signals-way/
- Lean Analytics: https://skills.wondel.ai/skills/lean-analytics/