The AI Builders Program

things you'll do

• Create the free-tier accounts you'll need throughout the program. Do it now, so live sessions stay focused on building — not waiting for someone's ‘Are you human?’ check.
• Email us three problem statements you would genuinely consider solving. For each, write one sentence naming the pain, who feels it, and whether you can reach five of those people within a week.

things you'll do

• Set up the tools you'll use throughout the program with mentors beside you, and understand the job each one does.
• Learn the ever-confusing tech lingo, so you can hold your own with the tech bros, follow what you're doing in the program, and explain what needs to happen next.
• Define the five tests the problem must pass. Use prompting to discover and synthesize, then learn where AI summaries stop being evidence.

things you'll do

• Define the user stories for your first target audience: what they need to do, where they hesitate, and what ‘done’ looks like.
• Define one killer flow in five steps or less. Everything else earns a place on the out-of-scope list before six weeks becomes six months.
• Turn the user stories and killer flow into a spec agents can build from. No clever prose. No expensive ambiguity.
• Learn the GitHub workflow and how Claude agents divide work and hand it back: branches, commits, reviews.
• Set the architecture, repo, and first slice: schema before screens, the boring stack, durable project memory, and a live scaffold.
• Use Claude Design to shape the flow before you spend build time polishing the wrong screen.

things you'll do

• Work on your design craft. Spot the AI tells that make interfaces generic, then replace them with choices that look intentional — not merely clean.
• Synthesize the best design decisions from five beautiful websites or products. Steal the thinking, not the pixels.
• Explore design components and skill libraries that raise the floor without turning your product into template soup.
• Learn agentic engineering: plan before write, build vertical slices, give subagents focused jobs, add reviews and tests, and keep Git in the loop.

things you'll do

• Learn the art of usability tests and user interviews: ask better questions and catch what people do — not just what they politely say.
• Run the first real-user loop with at least five people. Do not explain the product or interface; ask them to think aloud as they maneuver it. If you have to narrate it, the interface has already answered.
• Maintain the Keep/Kill/Change Log and turn observations into decisions. Kill items get deleted, not moved behind a flag to haunt the repo.
• Define the success metrics and set up the analytics tools that will capture them before launch. Vanity numbers can sit this one out.

things you'll do

• The hard 20%: make auth, RLS, and emails behave; migrate without losing data; pressure-test error states and edge cases; handle background jobs; and recognize the failure modes agent-written code prefers to hide.
• Prepare the QA checklist from real user stories: edge cases, failure paths, and everything that can go wrong — before a user finds it first.

things you'll do

• Finish production readiness: clean up secrets and environments, add rate limits, backups, and error tracking, then sharpen the positioning and build the launch assets. The last 20% is where demos become products.
• Connect your domain and give the product a real home. A preview URL proved the build works; this is the address you can put in public.
• Set up Google Analytics and submit the site to Google Search Console. Launch with signal, not a shrug.
• Run the final launch war-room before Demo Day: fix what breaks with mentors, then prepare the problem, the flow, the live product, and the numbers for the room.

things you'll do

• Each builder gets 15 minutes to show the problem, the flow, the live product, and the numbers available so far.
• After the Demo Day call, launch the product publicly and add it to the builder shelf. The live product — not the recording — is the portfolio asset.