Launch playbook

A short checklist for taking a Buildspace app from idea to real users. Each step links to the detailed docs. Run it once for launch, then keep looping steps 6–7.

1. Plan it

• Write down who it's for, the one job it does for them, and what "working" looks like.
• Save the plan in the repo (__plans__/1-<slug>.md) so you and your agent build from the same page.
• Keep v1 small: one core loop, and nothing you can't explain in a sentence.

2. Build it

• Create the app and hand your agent the context: see Build with an AI agent.
• Use the platform for auth, database, events, storage, email, and billing instead of adding your own.
• Ship to your dev preview often: git push, then buildspace agent reset.

3. QA it

• Run your checks before every push (lint, typecheck, build, tests).
• Click through the real flow on the dev preview as a new user, on a phone as well as a laptop.
• Check the basics people trust: Terms and Privacy pages, error states, and empty states (buildspace app status shows the legal checklist).

4. Ship it

• Promote to production: buildspace promote --latest --yes --watch.
• Point your own domain at it with Custom domains.
• Smoke-test what actually shipped: the health check, the home page, sign-in, and one full core-loop run.

5. Get found: SEO and AEO

People now find apps through search engines and through AI assistants (answer engines). Both reward the same things.

• Be crawlable: a sitemap, a robots.txt that allows public pages, a unique title, description, and canonical URL per page, and share images.
• Have one indexable host: make sure only your real domain is indexed, not your dev preview or fallback URL.
• Answer real questions: give each page one intent, put a short direct answer under each question heading, add comparison tables and FAQs, and only claim what ships.
• Be readable by agents: publish an llms.txt and markdown versions of key pages, and document any action an agent can take for a person. Anything an agent can't do itself should become a link the person opens and confirms.
• Tell the engines: verify the domain in Google Search Console and Bing Webmaster Tools, and submit your sitemap. Many AI assistants lean on these indexes.

6. Market it

• Start with the people who already have the problem: communities, newsletters, and your own network.
• Get listed where assistants look: directories, comparison and "alternatives" articles, and honest answers in forums like Reddit.
• Make every user a channel: shareable pages, share cards, and a clear "get your own" link.

7. Measure and iterate

• Track the core loop with events, and watch it with buildspace app insights or buildspace app report --json.
• Record where signups come from, including AI referrals.
• Each cycle, pick the biggest drop-off or the most-requested feature, write a short plan, and go back to step 2.