Action Plan Generator

Turn any YouTube video into a 7-day action plan

TypeScriptAstroVue 3OpenAI API
Role
Independent full-stack developer and product designer
Context
Video summaries explain content but rarely turn it into action
Implemented
A constrained LLM pipeline, structured plan UI, copy flow, and share links
Constraint
Keep the experience stateless, immediate, and specific rather than freeform
Outcome
A working tool that converts one video link into a time-boxed seven-day plan
Status
Live product prototype with public source

Watching self-improvement content rarely changes behavior, but acting on it does. Every AI tool in this space produces summaries, which just retell the video. This app produces something different: a concrete, dated, day-by-day plan you can start today.

The walkthrough below is a real run. I pasted Tim Urban’s TED talk "Inside the Mind of a Master Procrastinator" and captured what came back.

01

One input, one button, done

One input, one button, no login, no saved history. The value is in the output, so everything before that takes five seconds.

The Action Plan Generator landing page with a single YouTube link input
02

It extracts what the video actually argues

From the TED talk it pulled the core idea and the three insights that matter: the "Instant Gratification Monkey," the "Panic Monster," and deadline-driven urgency. It then converted each into an action with a time attached: block distractions for two 30-minute sessions today, set a deadline within 24 hours. Insights on the left, actions on the right; understanding versus doing.

Generated result from a TED talk: core idea, three insights, and three concrete actions
03

Seven days, one concrete task each

The plan is deliberately specific: "60 minutes before noon," "10am and 3pm," and "write 3 bullet points tonight." Vague plans are the problem this app avoids. A copy button and share link let you get it out of the browser before you watch another video.

A generated 7-day plan with one specific, time-boxed task per day
  • TypeScript, Astro server routes, and a Vue 3 interface, with an OpenAI Responses API pipeline that extracts the video’s content and reshapes it into the fixed plan structure: core idea, insights, actions, and a 7-day schedule.
  • Stateless by design: no accounts, no stored history. A generated plan lives in the share link.
  • Prompt-shaped output, not freeform: the model is constrained to produce actions with times and quantities, which is what separates a plan from a summary.

Like what you see? I'm available for full-stack and automation work.

Get in touch