Veritas Kanban, our new, open source agentic AI orchestrator went live today. Visualize and plan your agents flows! We have nine new features planned for v1.1, and would love for your feedback or assistance. All aboard the Clawdbot/Moltbot hype train!🚂 github.com/BradGroux/veritas…

Anthropic's motives are clear. They urged for regulation to stifle the competition, specifically open-source models. Local, open-source models, personal to you and your data, are the future. Don't sacrifice your tomorrow, for subsidized tokens today. Tell Dario to suck it.

Veritas Kanban v5.0 is live. I built Veritas around a simple belief: AI-assisted work needs an operating system, not just a chat window. Once agents start touching repos, writing code, running checks, creating PRs, generating docs, or preparing releases, you need visible task state, scoped permissions, approvals, evidence, audit trails, and recovery paths. Veritas started as a local-first Kanban board. v5 turns it into a desktop command center for human AI software work. VK was built by OpenClaw, for OpenClaw, but works with any agentic platform. What changed in v5: • Signed and notarized macOS desktop app (store app coming soon) • SQLite-backed storage with migration, backup, export/import, and recovery paths • Multi-user workspaces, roles, scoped API tokens, device sessions, and RBAC • Mobile/PWA access for trusted hosts • Task Work View, action queues, readiness gates, work products, and completion packets • Workflow authoring, policy decision traces, universal search, and Maintenance Center • Agent provider profiles for Codex, Ollama, LM Studio, and other local/server workflows • CLI, REST API, MCP, OpenClaw, Squad Chat, and workflow integration surfaces The important part: Veritas still starts simple. You can use it as a board-only local app with no agents, no cloud account, and no extra orchestration. Then you add OpenClaw, Codex, MCP, workflows, webhooks, remote access, or governance only when the workflow actually needs that layer. That matters because agentic work gets messy fast. The goal is not "more automation." The goal is automation you can inspect, constrain, review, repeat, and recover from. Install on macOS: brew tap BradGroux/tap brew install --cask veritas-kanban Repo: github.com/BradGroux/veritas…

Microsoft Build was one of those weeks I’ll be thinking about for a long time. I presented at Backstage at Build. Networked at GitHub Build on Tap. Two packed days in the Open-Source Zone. Eight straight hours talking with people about OpenClaw. A few "secret location" moments (more to come soon)! A couple of private dinners with very sharp people. Then closing it out with the amazing OpenClaw After Hours at GitHub event. I joked during the week that I was the dumbest person in the room, and I meant it as a compliment to the room. Huge thank you to @OpenClaw, @Microsoft, and @GitHub for making the week possible, and for giving this project and community such a strong place to show up. Special thanks to @ashleywolf, @KevinCrosby1, @GreggCochran, @OmarShahine, @shanselman, @steipete, @vincent_koc, @davemorin, @ArnoldCastro, @RegGroux, and the many others who made the week what it was. The best part wasn’t the schedule. It was the nearly 20 of my fellow OpenClaw maintainers, the people who stopped by, asked hard questions, shared ideas, challenged assumptions, and got excited about where OpenClaw is headed. GitHub Universe, anyone? #MicrosoftBuild #OpenClaw #GitHub #GitHubUniverse Thanks to all of my fellow maintainers: @BunsDev, @somalley108, @dallinfinite, Agustin Rivera, @pat_erichsen, @joshavant, @jlehman_, @_JacobTomlinson, @jesse_merhi, @thescottfan, Sarah Fortune, and @kevins8.

Monday is the last day to apply for our AI delivery partners and internship opportunities at @DigitalMeld. We don't care where you went to school, or where you've worked. We care about how you think, and what you're building. Apply today: sstb.ai/team.html

Such a privilege to work with Microsoft to bring claws to enterprises!

Go, Brad, Go! Presenting at the Backstage @ Build event. ✨ 🦞

IAH ➡️ SFO. #MicrosoftBuild is loading… going over my #OpenClaw in 10 Minutes presentation for later today at GitHub. Follow along: where.bradgroux.com

I built a little live presence page for my week at Microsoft Build: where.bradgroux.com (built with OpenClaw and Codex) Use it to see where I’ll be, what sessions/events I’m around, and how to find me between things. I’ll be using it to share check-ins from the OpenClaw journey at Build, with about 20 of my fellow maintainers. We'll be at several sessions and events with GitHub, Microsoft for Startups, at the Open Source Zone, and at @steipete's session. Come say hi! #MicrosoftBuild #GitHub #MicrosoftForStartups

AI is not eliminating the need for management. It is punishing vague management. Agents need clear priorities. • They need context. • They need constraints. • They need review standards. • They need someone to decide what matters when two goals conflict. That is management. The title may change. The work will not. Better managers will get more leverage. Vague managers will get louder chaos. Start small, think big.

Your SOPs are your software now. Not because documents are suddenly exciting. Because AI agents can only run the business process you can explain. • If the process lives in someone’s head, the agent guesses. • If the process has no examples, the agent improvises. • If the process has no review gate, the mistake ships. The companies that document well are about to automate faster. Boring work. Big advantage. Start small, think big.

The wrong question is: "What can AI do for my business?" That question is too big, too vague, and usually turns into tool shopping. Ask this instead: "What work happened three times last week that annoyed someone?" That’s where to start. Recurring annoying rules-based = automation candidate. Not everything needs an agent. Not everything needs a new app. Not everything needs a transformation project. Sometimes the win is one clean workflow that gives a person 30 minutes back every morning. Stack enough of those and the business changes. Start small, think big.

AI didn’t remove the work. It moved the bottleneck. Before: "Who has time to do this?" Now: "Who knows what good looks like?" That’s the real AI skill for small teams: • define the outcome • spot bad output fast • build repeatable workflows • keep humans in the high-judgment spots The winners won’t be the people using the most AI. They’ll be the people who can turn messy work into clear systems. Start small, think big.

AI didn’t replace your job. Someone who documents their process, measures the output, automates the repeatable parts, and shows up with receipts became harder to ignore. The advantage isn’t "using AI." The advantage is turning messy work into a system other people can trust using tools like OpenClaw or Codex. That’s the whole game. Start small, think big.

Most companies are building their AI roadmap backward. They start with: • Which model? • Which agent platform? • Which plugin? • Which chatbot? Start here instead: • What decision repeats every week? • What handoff breaks the most? • What exception costs real money? • What approval cannot be skipped? • What system is the source of truth? The model choice matters. But the workflow choice matters first. Start small, think big.

The AI agent is not the strategy. The strategy is deciding: • what work should move faster • what work needs approval • what data is safe to touch • what exceptions require a human • what "done" actually means An OpenClaw agent without those answers is just automation with confidence. Useful, maybe. Dangerous, definitely. Start small, think big.

OpenClaw just plugged into X, and now your own hardware gets the claws. 🦞 Bring your Grok, SuperGrok or X Premium subscription to your OpenClaw agent. Now even your personal agent is red-pilled and based. Get Grokked: x.ai/news/grok-openclaw

The expensive part of AI usually isn’t tokens. It’s workflow entropy: • Bad handoffs • Missing context • Unclear approvals • Duplicate data entry • No audit trail No one knows who owns the exception path. Agents, like OpenClaw, don’t magically fix that. They expose it faster. That’s why the real AI work starts before the model call. Start small, think big.

Your UI is not the product. Your schema is the product’s skeleton. If the relationships are missing, the app doesn’t have bones. Users don’t care that the page renders. They care that their work, access, history, and permissions make sense. Plan and then execute with OpenClaw.

Midmarket and Small businesses are about to get weirdly dangerous. Not because they have better AI. Because they can replace a bloated SaaS workflow in a weekend while the enterprise is still scheduling the vendor review meeting. Speed compounds.