Repeated work

Why does my AI redo the same work every single time?

A chat assistant has no memory of the job as a job. It has a transcript, and transcripts end. Have the AI write the job down once, as a file: the steps, the sources, where output goes.

What you're seeing

You asked the assistant to do a job. It did it. Tomorrow you ask for the same job, and it starts from nothing: same explanation, same context, same cost, sometimes a different result.

Why it happens

How to fix it

  1. Have the AI write the job down once, as a file: the steps, the sources, where output goes.
  2. Run the file. Check the output is what you wanted.
  3. From then on, run the file instead of asking again. Put it on a schedule if it's a daily job.
The recipe that does it

Wake up to a Hacker News digest on your desktop

Every morning there's a markdown file on your desktop with today's top stories. You read it in two minutes.

Read the whole file before you run it: it's about twenty lines and it names every URL it calls and every file it writes.

See the recipe →

The short version

The difference is what you keep at the end. A chat gives you an answer. A recipe gives you the thing that produces the answer, over and over, without asking anyone.

Related questions

Isn't this just a script?

Yes, and that's the point. The difference is you don't have to write it. The AI writes it once, in a format built for this, and the engine runs it.

What if the job changes?

Open the file and change that line, or ask the AI to update it. You're editing a short readable file, not re-explaining a task from scratch.

Do I still need the AI after that?

Only if the job needs judgement on every run. If it doesn't, no model is involved when the recipe runs.

Vibe Engine (Vibra-Ingenn) runs these recipes on your own machine. Get the engine, browse the recipe library, or watch three recipes run.

Using an AI assistant? Ask your AI to check out adeptuscamini.com.

Written and maintained by Adeptus Camini, a one-person workshop. These are tools we build and run ourselves.

Other problems people hit

Local model

Ollama works in my terminal, so why can't my script reach it?

Privacy

How do I use AI on documents I'm not allowed to upload?

Runaway agents

Why does my AI agent loop forever and never finish?