We’re flying out to attend the Texas Land Title Association’s annual conference to show real people how leveraging AI through https://t.co/dKcKX1GI1F can help make their workflows faster. Humans using it, not replaced, and always with a person in the loop.
I’ve been gradually migrating a few projects over to Grok Build via ACP as the primary pipeline. Soo many things I love about the xai agents and models. Speed is refreshing!
we are resetting rate limits so you all can try out Grok Build more!
we now ship with a new checkpoint of Grok Build and Composer 2.5
as always looking for critical feedback to improve
@wesbos Crazy fast and generally works well. They are shipping improvements fast which is nice as well. We are working with it pretty heavily where we can to keep up with momentum.
Introducing Agentic Hot Reload: Hot reload that will work automatically with every coding agent ⚡
When you prompt your agent to make a change, it will reload your running app and show you UI updates in real-time.
Grok Build has an auto-approve mode now, so you don’t have to constantly babysit the workflow while building
Just press Ctrl + O to enable Auto-Approve mode and Grok Build will continue executing actions automatically
It even clearly displays which mode you’re currently in
The developer experience on Grok Build keeps getting better incredibly fast
Why is it that I often have to fight with Claude’s auto classifier to get it to do what I want? Ultimate it relents. I understand it’s trying to guide the conversation and help make good decisions but maybe just ask me to confirm. I’m 52 years old. I don’t need an agentic parent.
We’ve added two security improvements to Claude Managed Agents.
Self-hosted sandboxes keep the agent’s execution environment in your infrastructure or with a managed sandbox provider.
MCP tunnels let the agent connect to services inside your security perimeter.
Just added our MCP connector to grok for Sluicebox and holy crap it’s fast! I’ve gotten used to the speed of Claude over the past year but tool calls and lookups rip on Grok with remote MCP!
Spent the weekend with Grok Build on a couple of side projects. Model calls and reasoning are insanely fast. One-shot feature implementations land more often than they should. Plan mode is solid. Decisions are good. Highly intelligent, if that's what we want to call it. Hooked it up to some of our local MCP tooling for task management, markdown extraction, knowledge base and that works really really well right out of the gate.
I've been a huge fan of xAI from the start. We've had Grok running in production for a while with backend services, mobile apps, phone IVRs, even a few games. Voice models doing TTS and STT on a few mobile apps and game stuff in MinorMUD. The voice models are awesome.
We're a 26-year-old dev shop with about 10 devs. Heavy Claude Code users. Most days it works really well and we lean on it constantly. A while back we built our own coding agent on top of our internal chatbot, modeled after Claude Code's workflow but running on Grok models so we could test that out. That also works pretty well, personally I love Grok reasoning.
xAI got a slightly late start on the dev tooling side, Claude Code had the edge for a while but it was always a matter of when, not if. I see how Elon attacks problems and once he builds momentum it’s like a freight train.
What's not there yet: long-running sessions and deeper orchestration. Everybody's into TUIs lately, and Grok Build's TUI is pretty and works well if you're sitting in one console but I run 20–25 agents at a time, across different projects, in tmux. A TUI doesn't slot into that the way a normal terminal client does. What I'd love is a JSON backend with hooks. Something I can drive programmatically and plug into the orchestrator we built for our own use.
Anyway, going to spend as much time on it as I can until it's our main process. Its pretty lean and mean already. Claude Code still has its place for now, but I'm super excited to move more workflow over to Grok Build
Lots of real work to catch up on this weekend but on the side going to spend some time with @grok cli building on https://t.co/gVaky1XrDx. Pretty heavy Claude code user but super excited to see how grok does.
@wholemars Not alone. TUIs are nice and all and have their place but one of the big advantages of CLI is script ability and friendliness with existing tools like screen and tmux. TUIs often break those rules.
An early beta of Grok Build, an agentic CLI for coding, building apps, and automating workflows is now available for SuperGrok Heavy subscribers.
Through this early beta, we will improve the model and product based on your feedback.
Try it at https://t.co/bpTHpjivWD
We're heading across the country to the NC Bar Real Property Conference! Excited to explore new opportunities of custom AI software for title attorneys!