Snapper

Grok @bot Pages just launched. 4 examples of what you can build: - Project overview dashboard - Coding Bot cost tracker - Kit newsletter analysis - Interactive Bot team map I like that it can pull data from your connected apps & personal computer to give you instant analysis and insights.

Grok @bot just launched Pages. Here's one example I built using the Kit newsletter connector. My Bot pulled in the campaign results, built an interactive dashboard, analysed which topics perform best, and recommended what to send next. Full walkthrough with 3 more examples pinned to my profile if you want to see more.

Once you're running more than a couple of Grok Bots, the hard part stops being the bots and becomes keeping track of them. My version: one coordinator bot owns a shared Notion board, the other bots report to it, and I get one weekday morning update instead of checking every chat. @buckleyplanet’s To Do setup is a neat take on the same problem.

Working on a video about Grok @bot’s new Pages feature. One Page I’ve built tracks my Codex spend straight from the logs on my bot's computer, and another pulls my newsletter stats from Kit and suggests what to send next. What other use cases do you want to see?

I'm 47% through my weekly Grok Bot limit with 6 days to go 😬 Where it went: • Auditing a WordPress site for a migration project • Reading X's open-source algorithm code • Building + editing a 40-question intake form (TypeForm connector) • Coordinating my coding bot on 18 fixes Anyone else burning through @bot usage limits?

I connected Codex to my Grok @bot's personal computer. Now my AI agent team takes a website change request, writes the code, and opens a PR with a Vercel preview for me to review. From feedback comment to live update in about 5 minutes. I just approve and merge. Full setup + real-time demo in the video 👇

Here's what my notifications look like when someone leaves a feedback comment on Vercel. My Grok @bot team picks it up, logs it, and has a Codex prompt waiting for my approval within minutes.

I’m building a Grok @bot team to run my website development workflow. I’ve got agents that: → Monitor feedback comments on Vercel → Turn requests into actionable tasks → Ask clarifying questions when needed → Manage projects through Notion → Prepare development instructions for Codex Now I’ve connected Codex CLI + GitHub to Grok Bot’s computer, so an agent can do the coding itself and open a draft PR for me to review. Video walkthrough coming soon.

Want to build a team of AI agents to run your business? In this article, I break down my Grok Bot web dev system, including agent roles, workflows, real instructions, and guardrails, so you can adapt the same approach to automate your own business. 👇

Grok Bot just added email. Your AI bot can now have its own inbox. I tested it with four emails: → A customer enquiry: flagged as high priority, with a drafted reply → A hosting receipt: logged as an expense to keep on record → An AI newsletter: marked as info only → A prompt injection email trying to override its instructions: flagged as suspicious and ignored And when I asked it to send the customer reply, it drafted the email and asked for my approval first. Full walkthrough 👇

I’ve just launched the SnapperAI model benchmark leaderboards 📊 Compare the latest AI models across coding, multi-turn, OpenClaw and Hermes, including passes, repairs, cost and wall time. Full leaderboards now live on my website 👇 snapperai.io/benchmarks/codi…

Grok 4.3 just became the strongest all round model in my benchmark set. Ran it alongside GPT-5.5, DeepSeek V4 Pro and Qwen 3.6 Max across coding, OpenClaw and Hermes. It was the only model from this update that held up across all three benchmark families. Full rankings →

First look at how Kimi K2.6 did on my Coding Benchmark and OpenClaw and Hermes runtime fit benchmark. Solid results! youtu.be/2tdanpYIefU

I tested Opus 4.7 across three custom-built benchmark: Coding, Multi-Turn tasks, and OpenClaw & Hermes runtime fit. Looks like a very strong model for OpenClaw, full results here: Opus 4.7 Ranked on OpenClaw, Hermes & Coding Benchmarks vs 12 Models youtu.be/xK7GlzppQfc

GLM-5.1 is the best model to use on Hermes, and ranks very highly on OpenClaw too. I built a new benchmark to test models on OpenClaw and Hermes, testing persistent memory, tool discipline, protocol compliance, and injection safety. I put 11 models through the benchmark, including GLM-5.1, GPT-5.4, Kimi K2.5, Grok 4.20, and more. Check out the results below! youtu.be/XKAjwLD7PPs

Recorded a video walkthrough of how to set up most of @bcherny's favourite features from this thread! youtu.be/liHxw3X8vdw

I benchmarked 8 models on real engineering tasks (bug fixes, refactors, migrations). GPT-5.3 Codex, Opus 4.6, Gemini 3.1 Pro, Kimi K2.5, Qwen 3.5 and more. Full rankings & results ↓ youtu.be/bjJCAYqL7yg