The coach that gets skipped first
A MIT camera guide that checks the step, and drops the live look when the key hits 60 a minute.
- Category
- AI
- Who it is for
- Someone learning a hands-on step from a video that cannot see them.
- Possible model
- Open source
- Difficulty
- Hard
- Inspired by
- FollowAlong · FollowAlong
- Original launch
- Watch the launch · Nov 5, 2026
Difficulty is an editorial estimate of technical and operational complexity, not a score.
Startup idea analysis by Startup Ideas. Last reviewed Nov 5, 2026. How we read a launch.
The startup idea
Unnati posted FollowAlong on November 5, with SKakade21. A how-to video does not know you did the step wrong. They built an AI that watches your hands, catches the mistake live, and tells you how to fix it. The repository is unnatibshah/followalong. The license is MIT, copyright 2026 Samruddhi Kakade and Unnati Shah. Muse Spark writes the steps and judges what geometry cannot. SAM 3.1 measures what is there. A shared limiter keeps every learner under one API key's 60 requests a minute. Checks and questions always get through, after a short wait if they must. The live coach's looks are the first to be skipped, and the server tells each learner. Demo videos use stock footage the uploader could not relicense. The clips are cut into steps and otherwise unchanged, credited under CC BY 3.0. The post does not name the 60-a-minute cap, or say the footage in the film is stock. This page did not open the camera.
The opportunity is the check before the next step, for a task a video cannot see. A smaller product is one procedure, a check that waits in line, and the live coach marked as the first thing dropped. A guide with no key and original footage is not this repo.
Do not hide the cap. Do not call the stock clips original.
The problem
A how-to video cannot see that you already did the step wrong.
Who would pay for this?
- Primary
- Someone learning a hands-on step from a video that cannot see them.
- Secondary
- A team sharing one API key who will accept that the live look drops first.
- Early adopter
- One procedure, one camera, the 60-a-minute line.
How could it make money?
Possible models. Nothing here is a claim about what FollowAlong charges.
- The app is MIT.
- The check uses a shared API key.
- The demo footage is CC BY stock, not a grant of the original video.
What would the MVP look like?
- One procedure.
- A camera check.
- The live coach marked droppable.
- Stock footage credited.
How to validate the idea
- Ask five people whether the live coach kept looking after the key hit the cap.
- If they say yes, they missed the skip.
- Do not relicense a stock clip as yours.
How to differentiate
- The check waits. The live look does not.
- One key, 60 a minute.
- The film credits the stock.
- Do not become the how-to site.
Original product launch
This launch is inspiration and evidence that someone shipped, not a partnership. The idea above is our reading of the broader opportunity. FollowAlong did not write it.
- Product
- FollowAlong
- Source
- FollowAlong
- Film
- FollowAlong, the live coach is the first thing dropped
- Posted
- Nov 5, 2026 · 2:19
Product that inspired this idea
Unnati's post says a how-to video does not know you did the step wrong, so she and SKakade21 built an AI that watches your hands, catches the mistake live, and tells you how to fix it. The repository is unnatibshah/followalong. The license is MIT, copyright 2026 Samruddhi Kakade and Unnati Shah. Muse Spark writes the steps and judges what geometry cannot. SAM 3.1 measures what is there. A shared limiter keeps every learner under one API key's 60 requests a minute. Checks and questions wait and still get through. The live coach's looks are the first to be skipped. Demo videos use stock footage the uploader could not relicense, cut into steps and left otherwise unchanged, under CC BY 3.0. The post does not name the 60-a-minute cap, or the stock footage. This page did not open the camera.
Posted by @notunosh. Watch the original launch. No separate product site is listed unless the post itself is the source.
- camera
- agents
Related startup ideas
- A coworker that stays on after the chat
An agent with its own computer that keeps a project page, and can be reached from the chat app a team already uses.
- A self-hosted assistant for personal apps
An assistant a person can host, pointed at the apps they already use, including computer use.
- A worker that lives in a folder
Give each worker a folder on the machine, a department, and a job, and say on the first screen whether it asks before it acts.
- Ask the computer, then approve the change
Put the agent on the desktop the person already uses, and stop it before it changes a file.
- Memory that survives the next coding session
Sell a memory layer that learns how one developer works, then carries that into the next agent run.
- The log that cannot prove who wrote it
An MIT ledger an agent must ask before a payment, that does not move the money and cannot force the question.
Looking for more ideas?
The full list is startup ideas for 2026, each one tied to a launch you can watch.
Browse startup ideas for 2026