SharpAI
/

Sharp AI

Search every FTC doc
at once.

Game Manual 0, the official FTC Docs, the SDK reference, Pedro Pathing, REV and Chief Delphi. Around 3,500 sections, so you can ask a question instead of opening ten tabs.

Ask a question Browse the index
Draw the route the code writes itself

Every answer points at a page.

From the actual docs,

with the section it came from,

Nothing in the index covers that.

or it tells you it doesn't know.

Draw a path. Take the Java.

Path planner

Drag waypoints on a 144" field, watch the robot run the route, and copy out a PathChain that uses the real BezierLine, BezierCurve and heading interpolation calls. It flags the degenerate control points Pedro's own docs warn about.

Open the planner

Browse by system

Questions people have already asked, grouped by the part of the robot they are about.

How it works

What happens when you ask

  1. Your question runs through a keyword search and a semantic search at the same time, and the two lists of results get merged.
  2. If nothing in the docs really matches, you get told that. The check runs on the search scores, before any model is called.
  3. The model only ever sees the sections that came back. It has to say where each claim came from, and it is told not to invent part numbers, gear ratios or rule numbers.
  4. Answers keep a permanent link, with the sections they used sitting next to them.