Improve Open Interpreter Performance with Spreadsheets

Does your AI agent lose focus during a multi-step task?

Here's an approach that helps Open Interpreter stay on track PLUS it speeds up prompt engineering iteration.

[Fermion.]

Very nice prompting tip.

I'm constantly arguing with people that say LLMs are overrated.

My retort is GIGO. _Perhaps it's not the model, but possibly a PICNIC (Problem In Chair Not In Computer) issue._

[dangreller]

Cool idea - Could also allow you to store different system prompts to give the interpreter different personalities, similar to Custom GPT’s

[avi]

Great idea. I was just thinking too of another column called "tool" (which is basically a python or shell script, or whatever makes sense) with the instructions: if you see a tool, look in the tools folder for the matching tool and use the data in context yo execute it, otherwise if tool isnt found, create the tool and the use it to complete the step. -- something like that should save on tokens for future operations not having to generate scripts etc yo handle the same simple operations over and over again.

[thesvenni]

Nice. Quick question, I have been trying to load a spreadsheet with a business model I created for one business, and having Open Interpreter adapt it to a new business model using the same structure. Unsuccessfully unfortunately for now, any tips on how to do it correctly?