groby
13
2
groby has not written any posts yet.
A useful technique to experiment with if you care about token counts is asking the LLM to shorten the prompt in a meaning-preserving way. (Experiment. Results, like all LLM results, are varied). I don't think I've seen it in the comments yet, apologies if it's a duplicate.
As an example, I've taken the prompt Neil shared and shortened it - transcript: https://chatgpt.com/share/683b230e-0e28-800b-8e01-823a72bd004b
1.5k words/2k tokens down to 350 tokens. It seems to produce reasonably similar results to the original, though Neil might be a better judge of that. I'd have tried it on my own prompt, but I've long found that the value I derive from system prompts is limited for what I do. (Not impugning Croissanthology here - merely a facet of how my brain works)
Having been at the same conference: The gap was staggering. On the one side, people & teams who by now have deeply agentic workflows (not just code - individualized workflows across the board were a thing). In the middle serious discussions on rather useless things like "LOC isn't a good productivity metric, how do we count productivity now". And a large chunk just very disconnected from what is and isn't possible, in both directions.
Even if we're at the top of the S curve (personal take: Probably, at least without fundamental breakthroughs beyond "scale"), there's massive changes already deeply baked in, and for the unaware teams, it will feel like continued exponential growth... (read more)