Once you see how much crap they’re running to police the agents on the repo, you’ll ‘get’ the spend https://x.com/steipete/status/2055405041843052792
I won’t lie, if I had the access to this, I’d do the same exact thing.
I won’t lie, if I had the access to this, I’d do the same exact thing.
Do existing companies run entire end-to-end product integration tests on every single change they make to a repo to make sure something hasn't broken? No, they just architect things in a way such that a minor change to something can be tested in isolation. And that can be automated, deterministically and efficiently.
Where I work we can release changes to our production site in minutes almost completely autonomously with high confidence with absolutely zero AI agents in the loop. How did we do it? With lessons learned from the past 5 decades of professional software development experience.
Lets not forget what OpenClaw is at it's core. It's a glorified cron scheduler. Why on earth does any of this effort need to exist. It's not that deep, it's not that complex, it's all AI for AI's sake.
I run it in a firewalled VM and am very conscious about any tokens I give it access to - so far for all I know this was unnecessary.
PS. for me the core feature of OpenClaw isn't the cron, though that is nice. It's the memory and instant extensibility. Like it takes 5-15 minutes to add an SSH tool where all agent requests go through a manual review, together with a good auto loaded description that just works in all future sessions.
This is clearly an implementation and not a conceptual issue, as I had none of these issues using the same model with Hermes, for example.
Yes, that is _exactly_ the problem that is being solved. Is it easier to spin up some LLMs or pay a team of experienced engineers?
As inference costs fall, which will be cheaper?
He has a different opinion of what it means to be lean than almost everyone else. That's fine, he's allowed to, but it's something you have to understand to make sense of any of his comments on things. He has a radically different set of values to most people.
The execution in case of Openclaw is a hot mess.
If these methods prove successful it isn't going to matter. A user doesn't care if code is 'slop' or artisanal, so long as the app/site/whatever works.
If you can combine autonomous flows (and millions of dollars in tokens) to produce work comparable to a traditional engineering team, then why would the user care which wrote the app/site/whatever?
Agricultural mechanisation didn't eliminate human labor over the 20th century. A huge fraction of the world's farmers have little or no mechanization today, well over a century after the invention of the revolutionary farm tractor.
With apologies to Ada Lovelace, but humanity has been writing code in anger for only like, 80 years? We'll still be at it in a 100 more.
I'm personally just impressed with the rate of improvement and _hope_ that it will continue, and that inference prices will fall (or on-device LLM become more feasible/powerful).
Anyway I appreciate your perspective even though I don't necessarily share it
I don't think there's any way most people would call that lean. It's lean in exactly 1 axis which is people, but no one really cares about that, people is always a proxy for cost.
The site say 1200 Github contributors and looking on Github there are now 2105 so it doesn't seem to be dropping that much.
How would we build software in the future if tokens don't matter?
[...]
All that automation allows us to run this project extremely lean.
Good thing we cleared that up. Another gem from the "if we had self-driving cars, we could just have them cruise on the roads endlessly when not in use and get rid of so much parking space!" school of resource management...