LaunchChair is a spec-driven planning workflow for founders and AI builders who build with coding agents such as Codex or Claude. It is not a hosted one-prompt app builder. It sits around the coding agent and supplies research, product decisions, a living spec, build cards, and launch assets.
It offers two project types. Product projects take a new idea from validation through launch. Feature projects research, define, and build inside an existing product, and both produce agent-ready work.
How does LaunchChair work?
A phase-based prompt engine compiles each prompt from the current phase, bucket rules, allowed write paths, required JSON shape, downstream dependencies, and the project state already saved. Structured output is validated and written back into the living spec.
The loop reads project state, queues the next action, runs an agent, applies the output, and validates it. Failed checks are retried or remediated, and the loop pauses for human checkpoints such as SQL actions, account authorization, and manual QA.
Key features of LaunchChair
Research covers ICP pain, substitute behaviors, competitor gaps, and wedge choice. The living spec holds positioning, pricing, scope, screens, flows, and data models. Build cards carry acceptance criteria and verification steps, with consolidated SQL checkpoints on the Feature track.
Agents connect through an Agent API with scoped tokens, a protected stdio MCP bridge, or a guided CLI setup. The same spec also feeds landing copy, schema, llms files, SEO, GEO, AEO, outbound sequences, and lead and signal tracking.
Key Features
Market mapping — Researches ICP pain, substitute behaviors, competitor gaps, and wedge strength, and turns the raw idea into validation tests and a market wedge direction.
Living spec — Keeps positioning, pricing, scope, screens, flows, and data models in one spec that is reused across research, build, and launch.
Prompt engine — Compiles each prompt from the current phase, allowed write paths, required JSON shape, downstream dependencies, and saved project state.
Feature projects — Validates demand inside an existing product, writes a focused PRD with acceptance criteria, and generates dependency-aware build cards for the existing codebase.
Agent loop with remediation — Queues the next action, runs an agent, validates structured output, retries or remediates failed checks, and pauses for SQL, account authorization, or manual QA.
Agent API and MCP bridge — Scoped tokens let agents read specs, claim and complete runs, and export phase summaries. A protected stdio bridge connects local agents.
Guided CLI setup — Install @launchchair/agent and run launchchair setup codex or launchchair setup claude to handle login, AGENTS.md, MCP config, bridge startup, and connection checks.
Private starter boilerplate — Creates a private GitHub repo from a Next.js and Supabase boilerplate with authentication, protected routes, UI components, SEO scaffolding, and agent instructions.
Spec-derived launch assets — Generates landing copy, CTA strategy, schema, llms files, SEO, GEO, and AEO surfaces, launch plans, outbound sequences, campaigns, leads, and signal tracking.
White-labeled summary pages — On the Agency plan, shareable summary pages can be white-labeled for client work.
Pros & Cons
Pros
Keeps research, decisions, and acceptance criteria attached to every agent task, so context does not need to be restated.
Separate Feature track validates and scopes work inside an existing product without forcing a new stack.
Works with several coding agents through an Agent API, MCP bridge, and guided CLI setup.
All plans include unlimited product and feature projects and the starter boilerplate.
Pauses for human checkpoints such as SQL actions, account authorization, and manual QA.
Reuses the same spec for landing copy, SEO, GEO, AEO, and outreach.
Cons
The sources show no clear free plan or trial terms, only a Start free button without details.
The token and time savings figures are described as an illustrative model, and actual results vary.
It is not a hosted app builder, so you still need a coding agent to generate the app.
The Founder plan includes only 1 seat, and full team use requires the higher tiers.
Pricing
Founder $19 month
Builder $39 month
Agency $99 month
Frequently Asked Questions
What is LaunchChair?
LaunchChair is a spec-driven Product and Feature workflow for founders and AI builders. Product projects take a new idea from validation through launch, and Feature projects research, define, and build inside an existing product. Both produce a living spec, build cards, and agent-ready work.
How much does LaunchChair cost?
LaunchChair costs $19 per month for Founder, $39 per month for Builder, and $99 per month for Agency. All plans include unlimited product and feature projects. The site states there are no hidden LaunchChair token fees.
Does LaunchChair have a free plan or free trial?
The pricing section lists only three paid monthly plans. The site shows a Start free button, but the sources do not explain what it includes or how long it lasts. Check the site for current terms.
Is LaunchChair an AI app builder?
No. LaunchChair describes itself as the product workflow and living spec layer around coding agents, not a hosted one-prompt app builder. It helps validate the product, scope the MVP, create build cards, and give agents structured prompts.
Which coding agents work with LaunchChair?
The site names Codex, Claude, Claude Code, Cursor, Hermes, Grok, Kimi, Gemini, and other local agents. They connect through the Agent API, the MCP bridge, or the guided CLI setup.
Can agents run a whole project automatically?
Agents can operate the loop through the API or MCP bridge. LaunchChair pauses for human checkpoints such as SQL actions, external account authorization, and manual QA.
Can LaunchChair be used for an existing product?
Yes. The Feature project type validates demand inside an existing product, defines a PRD, and generates build cards for the existing codebase. It does not require a new stack or architecture.
Is LaunchChair suitable for agencies?
The Agency plan costs $99 per month and includes 10 seats, unlimited full-workflow Agent API and MCP use, and white-labeled, shareable summary pages. It is described as being for agencies shipping products and features for clients.