Awesome innovation from the @trycua team! We dont need to burn big model inference just to decide what element of a UI to interact with. Using a local small specialized model to handle execution is the way. Cua-S1 is applying almost the same philosophy as Jev, one layer further down. • screen + candidate actions ↓ S1 ↓ • element • action • abstain

TIL something really useful in Firstmate by @kunchenguid. I was working in a completely separate Grok build session, outside the Firstmate repo, and asked: “What is the most effective way to hand this work off to my Firstmate orchestrator?” It went digging and found `fm-inbox`. So apparently there is no need to wait for the Firstmate chat to free up or manually copy everything over. From another agent/session: `Firstmate/bin/fm-inbox.sh note "your message"` That writes a durable note to disk and pings Firstmate to pick it up on its next check. That means another agent can basically walk up to the orchestrator and leave work on its desk while it’s busy. Even better for larger tasks: The note is the doorbell. The file is the brief. Put the full handoff packet in a file, then send Firstmate a short note pointing to it. Now Grok, Codex, Claude, etc, can finish a chunk of work and hand the next step to Firstmate without needing the orchestrator session itself to be available. That is way more useful than I realized. Also in the same binary: - `list` - see what’s waiting - `status` - see what’s happening without interrupting - `ask` - side-question a model without turning it into fleet work - `say` - speech/output if configured Yet another Firstmate feature that shows why it’s an actual agent distro: vendor-agnostic coordination infrastructure with durable handoffs between independent sessions - no human meat proxy required. Firstmate repo:

🥧 Introducing GooeyPi - A GUI for the Pi Family of Agents. Supports: Pi, Oh-My-Pi (OMP), and Prime Agent Very excited to share this with you all. Being into local models, I wanted to have a separate harness for my local models so I don't have to constantly switch configs. And I didn't want to go back to the TUI (nerds). But, I didn't want to miss out on all the features that I know and love. Turns out, I didn't have to. I just needed to build it myself. So I did. Features include: > Fully featured GUI experience > Agentic Browser > Realtime Voice Agent > Voice transcription (local or API) > Computer Use > Automation Desk > Agent to Agent messaging > Ask Question Tool > Git Control > Terminal > Pets > and more I tried to capture all of the most popular features I could in this first pass, but we're only just getting started. Available in BETA on MacOS, Linux, and Windows. I need testers! I have only tested on Mac and Linux. If you want to help me with this project, please reach out. Available:

introducing Gemini 3.7 Flash: our most intelligent workhorse model yet for coding and agents this release comes just three weeks after Gemini 3.6 Flash, and is a direct result of developer feedback and algorithmic innovations that we look forward to bringing to future models 3.7 Flash delivers substantial improvements across software engineering, knowledge work, and web development workflows — with an introductory price of half the original 3.6 Flash cost per million tokens

@bot still has no Android app, so I built a self-hosted Telegram bridge. Text, photos, voice, files, desktop mirroring, commands and Approve once / Deny Runs on the Grok computer.

@bot still has no Android app, so I built a self-hosted Telegram bridge. Text, photos, voice, files, desktop mirroring, commands and Approve once / Deny Runs on the Grok computer.

Compound Engineering Update Gotta say, @trevin and @kieranklaassen cooked on the updates to Compound Engineering! I daily drive Codex desktop app and in the past with 5.6 Sol I had to constantly remind it to use subagents, not do all the work itself. The new plugin is first class. Codex uses all the skills automatically with no reminder, in the correct progression and spins up subagents consistently. And the finished work is high quality engineering. Highly recommend!

@cognition Grok 4.7 will exceed all current models. That said, Anthropic is a great company and will probably release improved models soon. However, the SpaceX training corpus is so awesome & unique that I would be shocked if any model is better at real-world engineering than 4.7.