@ay_ushr agreed, and what has helped me immensely are @capydotai captains, with workers setup in cloud, and applying @potetoโs inverse pyramid framework. makes it feel more single threaded again
donโt wait for the agent, make the agent wait for you
We recently implemented this at Rocket and its hard to overstate how well it works, worth a quick swap in to test out. Great work @_aj and @MiguelriosEN
What if your agents could improve themselves and use fewer tokens as they run?
The labs aren't incentivized to do this, so we built ourselves, and now we're open-sourcing our entire self-improving agent platform and control plane.
And it works with Claude/Codex/Pi agents.
Fewer tokens, same work. Particle-1 from @particle_api uses 28โ52% fewer output tokens than DeepSeek V4.1 Flash across four coding benchmarks, per Particle. Now on OpenRouter.
Today is full of releases of insane optimism and optimization.
Suhail has the track record to build fantastic products.
I believe 2027 will be the year we keep twisting our heads basically daily.
weโre just getting warmed up.
1/ Excited to announce our new company Particle
This is our first step in a series of models focused on developers.
With particle-1, we focused on token efficiency and managed to keep parity in performance while using ~45% fewer tokens.
Live now: https://t.co/m5MvOwR9yE
why do llms write โcolourโ i would assume more examples of โcolorโ exist in the pretraining corpus. something in RL/post changes it? something else?
ok, after a few weeks I can confidently say @capydotai has completely changed how I work for the better. I'm now able to stay on top of 5-10+ things going on in parallel, something I've always struggled with before
contrib graph now looks like I've done jack shit before, but its just normalized
thanks for the tip @garrytan