Thanks for the question! Similar to another user’s comment about Paperclip, Herdr looks to provide a layer on top of existing agent harnesses, whereas with Pizza Bot, we run the execution loops for each agent ourselves (built on LangGraph / DeepAgents). This gives us control over checkpointing, human-in-the-loop, etc., and the only “dependency” is your model provider.
I’ve been describing Pizza Bot as an “inbox” because it captures the UX, but this does seem to cause confusion and undercuts the fact that Pizza Bot actually is an agent harness itself, not just an orchestrator of agents hosted elsewhere.
So far claude has been fine to generated zig code. Dont have lots of issues. Tijy local models have more problems with zig, need additional instructions.
In general, what helps is to instruct ai to use test coverage to ensure the cover all edge cases with tests.
For me, being quite "dyslectic" - LLM generated sentences have the advantage of being spelled correctly and being grammartically correct.
I rather prefer iterating over LLM text than writing it by myself.
Example: Typing this simple text above by myself was not straight forward.
Agreed that’s a good practice but for some very hardline people even that’s not considered acceptable. It’s very hard to understand what people consider right and wrong so I’ve just stopped listening.
No tolerance policy for 77 of 10k lines, but exceptions made for language with unashamed use of large swaths of LLM code because it’s popular with the author. People just aren’t being consistent at all. Even when they draw hard lines.
Well, those people would be opposed to using spellcheck and grammar tools, too, I guess. I don’t think there’s a reasonable argument against using AI to spellcheck/check your grammar other than that it’s hugely inefficient.
The argument as I've heard it explained to me by someone with a banhammer: "One use of AI no matter how benign is not acceptable because usage is infectious; after even the most minimal use of AI, we can't tell what was AI and what wasn't, therefore the only acceptable position is no AI whatsoever"
Surely we can all agree that banhammer people are generally not a good standard to go by? Still, if you value participating in that space, you have no choice but to follow their rules
Sure. For small changes it would work. However, for larger documents the effort increases a lot. LLM speeds things like "making use of terminolgy" in the document consitent.
Restructuring, combine parapgraphs etc. a lot faster.
In the old days it would take me weeks to get a good document out of door. Now a it is days.
While the absolute number of iterations polishing is higher.
reply