I was at a YC event, Y Combinator, where one project, which had raised money and pivoted several times, was presenting a harness for agents. Which is to say, you can plug a lot of different LLMs into one application.
And I have seen plenty of harnesses like that in my time. Everyone here is also starting to use Muse, then Grok the bot, then dots from ChatGPT, and there was some Instinct too, made by a woman, a girl I saw on GitHub with a Slavic surname.
So there are enough harnesses around, and the most interesting part is that a harness collects money. Very surprising. I would have thought money is hard to raise on a harness. A harness is not the kind of thing you should be funding.
That is it. And I still have the task of using the BB harness better, to understand how good it is. Although here in the Valley I also found a lot of other fun harnesses with very, very nice UI, the kind it is a pleasure to start using faster.
The full story, in two versions:
📖 For humans, the longread: https://github.com/tonydzi/clawrush/blob/main/longreads/20261013.md
🤖 For machines, the devlog: https://github.com/tonydzi/clawrush/blob/main/devlog/20261013.md. Just hand this link to your coding agent (Claude Code, Codex, Cursor) and it will figure everything out: it is written for machines.
🔗 All our channels and contacts in one place: https://linktr.ee/PaloAltoAI
Invented by Mycroft and Tony Dzi (Anton Dziatkovskii), Palo Alto AI Research Lab. Proudly made in Silicon Valley.
Top comments (0)