Good ChatGPT experience — was creating a list of categories of software engineering concerns. I fed that into chatgpt and told it to expand it and add more subcategories and it did a bang up job. Results get a lot better if you provide a more detailed initial template

Tech entrepreneur. Founder @electricsql. Founder @gatsbyjs (acquired by Netlify). @kylemathews on Twitter

I’ve used it in a similar way as an input for building a medium resolution map of B2B software capabilities. Medium res being that yucky middle ground between high level/general summaries and maps that are too tightly coupled to disparate systems.

in motion | founder, researcher, scout, storyteller | into computing, communication, complex systems | rewilding the web with subset.network | hq: msweet.net

I expect so! AFK/mobile atm but I’ll poke around and come back 🙃 did notice, during the exercise, that ChatGPT seems to get distracted when I try to prompt intense enumeration. Often would duplicating nested items under different parents or go completely AWOL and disregard requested structures or schema.

Had a look but nothing tangible I can drop here. Perhaps the most interesting finding for me is that LLMs are necessary but not sufficient for building intermediate resolution knowledge graphs (1,000s to 10,000 nodes).

interesting! Do you have some examples you can share of what you built?