Basis Lab

Putting a model into service

Build a system where a model decides to call something outside itself, run that loop safely, and know exactly which failures belong to the model, which belong to the tools, and which belong to the loop you wrote around them

Models That Use Tools

Giving a model the ability to do things rather than only say them. The loop, the tool descriptions, and the long list of ways the loop goes wrong.

8 lessons, written and corrected before you arrived. Reading them here needs no account. The first reads the whole way through; the others open and then stop, because a page nobody owns cannot tell who is reading it. Starting the course gives you your own copy, where every idea has problems standing under it and you can ask about any sentence.

Start reading

This is course 3 of 9 in a reading track: put a model in front of people and know whether it is working. The track is an order, not a gate, and the rest of it is on one page.

  1. 01Saying Something and Doing SomethingA model never does anything. It produces text, and a program you wrote decides whether that text becomes an action, which is where the whole subject lives.
  2. 02The Loopopening onlyThe whole of a tool using system is a short cycle: ask, call, return the result, ask again. What stops it is the interesting part, because nothing stops it by itself.
  3. 03Describing a Tool Wellopening onlyThe description is the only thing the model knows about a tool, so a vague one does not produce hesitation. It produces a confident call with the wrong values.
  4. 04Checking What the Model Asked Foropening onlyThe program should treat a requested call the way it treats a form submitted by a stranger: check the shape, the values and the permission, and refuse in a way the model can recover from.
  5. 05When a Tool Failsopening onlyTools time out, refuse and return nothing. The question is which failures are worth another attempt, and what the model should be told about the ones that are not.
  6. 06Several Steps and a Budgetopening onlyA loop with no limit is not a capability, it is an invoice. Bounding the steps, the time and the money is what makes a multi step system safe to run.
  7. 07Text From Outside Is Not an Instructionopening onlyA page the system fetched can contain a sentence addressed to the model. The repair is not better detection, it is arranging things so that fetched text cannot authorise anything.
  8. 08Judging Whether the Whole Thing Workedopening onlyA tool using system has to be judged on finished tasks, not on individual replies, and every failure has to be attributed to the model, a tool or the loop you wrote.