It's like writing a novel in a week that no one wants to read. If in six months you can do it in an hour, there is still zero value.
I haven’t seen credible compiler engineers engage with it. Look at the repo: issues and pull requests look like spam.
Finally, if you read the description of the techniques used in the parallel agent harness: there are no novel ideas.
I mean the blog acknowledges this with things like: "This is a very early research prototype. I haven’t yet implemented any other method for communication between agents, nor do I enforce any process for managing high-level goals".
But this is 2026, and the company is on its way to a trillion dollar valuation. Maybe research some distributed systems first and then try to apply those in a way that is a unique fit for agents.
Couple of obvious ideas: use a CRDT, use some existing lightweight synchronization algorithm like Lamport's bakery. Do some initial TLA+ modeling. Anything really. Just show us something better than slop at the level of the harness and that of the code output.
One year old example of an impressive and useful agent harness: "TxGemma: Efficient and Agentic LLMs for Therapeutics" from https://deepmind.google/research/publications/153799/
Let me explain why:
> the resulting compiled output is over 60kb, far exceeding the 32k code limit enforced by Linux
Seems like a failure to me.
> I tried (hard!) to fix several of the above limitations but wasn’t fully successful. New features and bugfixes frequently broke existing functionality.
This has code smell written all over it.
----
Conclusion: this cost 20k to build, not taking into account the money spent on training the model. How much would you pay for this software? Zero.
The reality is that LLM are up there with SQL and ROR(or above) in terms of changing how people write software and interact with data. That's a big deal, but not enough to support trillion dollar valuations.
So you get things like this project, which are just about driving a certain narrative.