Welcome, humans

OpenAI researcher Noam Brown dropped a fascinating essay asking a question we don't hear enough: who actually gets to decide what AI is allowed to do?

His point: right now, the rules around AI surveillance and autonomous weapons are being hammered out through procurement contracts and company usage policies. Not laws. Not courts. The Anthropic-Pentagon standoff is a perfect example; Anthropic drew two red lines (no mass surveillance, no autonomous weapons), the Pentagon wanted unrestricted access, and we ended up with a supply chain risk designation instead of, y'know, an actual law about it. Somewhere, a congressional aide is Googling "what is a chatbot."

Brown argues the fix is straightforward: Congress should legislate when AI can track Americans, close the loophole on purchasing commercial data, and codify rules around autonomous weapons instead of leaving them as Pentagon policy that can change with any new administration. His closer nails it: "AI will not be limited, in its consequences, to one company or one administration. It will change the lives of everyone. The laws that govern it should reflect that fact."

Here's what happened in AI today:

• 🙀 A new open-source tool lets you spin up a full AI-run company with org charts, budgets, and governance in one command.

• 📰 Microsoft launched Copilot Cowork, built on Anthropic's Claude, to complete multi-step tasks across Office apps.

• 📰 OpenAI is acquiring Promptfoo, an AI security testing platform used by 25% of the Fortune 500.

• 🧪 Anthropic sued the Pentagon over its "supply chain risk" label, calling it "unprecedented and unlawful."

• 💡 Learn how to use Microsoft’s new Copilot Cowork in Frontier.

🙀 In a world where China (and pretty much everyone else) is wild about OpenClaw… meet PaperClip

OpenClaw mania is taking off in a big way in China; the github repo has received more stars than Linux, and meetups have made way to a full on OpenClaw store in China. To say people are excited about it is an understatement. 

While Claw-mania consumes the consumer, it's the revenge of enterprise AI this week; and to say everyone wants their own version of OpenClaw for the enterprise is an understatement. 

First of all: Microsoft just launched Copilot Cowork, a feature built in collaboration with Anthropic that lets AI complete multi-step tasks across Outlook, Teams, Excel, and PowerPoint. Remember when Claude Cowork wiped $220B off Microsoft's market cap in January? Microsoft's response: take the name, license the technology, and ship it as a Copilot feature. The ultimate "if you can't beat 'em" move. (We wrote about Microsoft's enterprise security play with Agent 365 just this morning if you want the deeper context.)

Meanwhile, Anthropic launched Claude Marketplace, a commission-free app store for enterprise tools built on Claude. NVIDIA is also planning to launch an open-source AI agent platform, which they’re calling NemoClaw. The company has a NeMo Agent Toolkit, an open-source library for connecting, evaluating, and accelerating teams of AI agents across any framework. Jensen Huang called OpenClaw "probably the single most important release of software ever”... so it makes sense to try to package it up themselves.

Missing in action? Google still hasn't shipped its answer to Cowork. We also heard a rumor that OpenAI might ship its own cloud-based work tool sometime in the next few weeks as well… The message is clear: the "AI as your coworker" race is on, and enterprises are the prize. 

But the most interesting entry isn't from Big Tech at all.

Meet Paperclip (code): an open-source tool that organizes your AI agents (Claude Code, OpenClaw, Codex, Cursor, whatever) into an actual company structure. As creator @dotta puts it: "You can only manage a rats nest of shell scripts and HEARTBEATS.md for so long before you realize there's got to be a better way."

Context: OpenClaw (the popular open-source AI coding agent) uses a HEARTBEATS.md file to track tasks. Run multiple agents and you get a mess of terminals with zero coordination. It's like running a company where every employee works in a different building.

Paperclip gives you:

• Org charts and roles: AI CEO delegates to CTO, who delegates to engineers. Real hierarchy.

• Goal alignment: Every task traces to a company mission, so agents know why they're working.

• Cost control: Monthly budgets per agent. Hit the limit? They stop. No $500 surprise bills.

• Heartbeats: Agents wake on a schedule, check work, act. No babysitting.

• Full audit trail: Every decision, tool call, and conversation logged.

Want to try it? Here's how (no coding required):

1. Install Node.js (version 20 or higher) if you don't have it (ask your AI to help).

2. Open your terminal (Mac: search "Terminal" in Spotlight; Windows: search "Command Prompt").

3. Type npx paperclipai onboard --yes and hit Enter.

4. The app launches in your browser and walks you through setting up your first AI CEO.

5. Approve hiring a coder. That's it. Your "company" is running.

The project already has 14.2K GitHub stars and 1.6K forks just in its first week. A marketplace (Clipmart) is coming where you'll download entire pre-built company templates (content agencies, trading desks, dev shops) and run them with one click.

Why this matters: 2025 was the year of the AI employee. 2026 is shaping up to be the year of the AI company. Microsoft, Anthropic, OpenAI, and maybe now NVIDIA are all fighting over enterprise contracts. But Open-source projects like Paperclip are asking a different question: what if you skip the enterprise vendor entirely and build the company yourself? And by “yourself”, we mean by agents, of course.

And yes, the name is a Clippy joke. We think. We hope.

FROM OUR PARTNERS

Meet Slackbot, your personal AI work agent

Slack was already a major productivity enhancer, but the new Slackbot changes everything. 

Slackbot is a personal agent built directly into your Slack workspace. Slackbot handles the busywork by understanding your work, collaborators, and communication style, letting you focus on high-priority tasks.

• Search messages, files, people, and workflow history.

• Chat with Slackbot just like you would a teammate.

• Get all the power of AI with enterprise-grade security .

  The best part? You don’t need any experience with agents or even AI to use Slackbot.

Read the guide today. 

AI Skill of the Day

Copilot Cowork is the new "agentic" layer in Microsoft 365 that moves beyond chatting with Copilot to actually delegating work. Instead of asking Copilot a question and getting a response, you describe an outcome and Cowork builds a plan, executes across your apps, and checks in at clear milestones for approval.

Here are four things you can delegate to Cowork right now (if you're in the Frontier program):

• Calendar triage: Cowork reviews your Outlook schedule, flags conflicts and low-value meetings, proposes changes, and executes them once you approve.

• Meeting prep: It pulls together a briefing doc, supporting analysis, client deck, and pre-read email from your emails, files, and past meetings.

• Customer follow-up: After a trip or event, it compiles notes across emails and meetings, drafts a summary, and sends personalized follow-ups.

• Product launch planning: It builds a competitive comparison in Excel, drafts a value proposition doc, generates a pitch deck, and outlines milestones.

The key difference from regular Copilot: Cowork runs in the background across multiple apps at once, powered by what Microsoft calls "Work IQ" (its understanding of your emails, meetings, files, and work relationships). You stay in control with approval checkpoints.

Try this prompt (once you have access):

Prepare for my meeting with [client name] on [date]. Pull together a briefing doc from our last three email threads, create a one-page summary of their account status from my Excel files, and draft a short agenda. Send me a checkpoint before you email anything.

Cowork is in research preview now and rolling out to the Frontier program in late March. The new $99/month E7 tier includes it, along with the new Agent 365 platform for managing AI agents across your org.

Favorite insight: Cowork uses the same agentic engine as Anthropic's Claude Cowork, but wired into your entire Microsoft 365 data graph. Think of it as Claude Cowork with a corporate badge and access to everyone's calendars. For a deeper look at how Microsoft is locking down agent security, check out our explainer on Agent 365.

Want more tips like this? Check out our AI Skill of the Day Digest for this month.

Have a specific skill you want to learn? Request it here. 

Treats to Try

*Asterisk = from our partners (only the first one!). Advertise to 675K+ readers here!

1. *IronClaw is an open-source alternative to OpenClaw built to run always-on AI agents with hardware-enforced security. Deploy your secure agents now at ironclaw.com

2. Paperclip orchestrates a team of AI agents into an actual company with org charts, budgets, and governance; open source, self-hosted, free to run. (GitHub)

3. GitClaw runs your AI agent inside a git repo where identity, rules, memory, tools, and skills are all version-controlled files you can fork, branch, and audit; free and open source.

4. OpenMatter builds infrastructure for real-world agents, exposing six primitives (identity, storage, place, capability, policy, context) so AI can book couriers, resolve addresses, and execute physical-world actions (launch thread).

5. context-hub from Andrew Ng gives coding agents reliable, versioned, searchable API documentation in markdown so they fetch accurate context instead of hallucinating from web searches.

Around the Horn

• OpenAI is acquiring Promptfoo, an open-source AI security platform. Think of it as a "stress test" for AI apps: it automatically tries to trick, jailbreak, and break your AI system before you ship it, catching vulnerabilities like data leaks and prompt injections. Used by 25%+ of Fortune 500 companies.

• Microsoft also announced a new $99/month E7 tier for Microsoft 365 that bundles Copilot, agent management, and identity tools. That's 65% more than the current E5 plan.

• Harvey introduced Agent Builder, letting legal teams build, share, and schedule autonomous agents for complex work like due diligence reviews and regulatory monitoring with human-in-the-loop checkpoints.

• Figure demoed Helix 02 tidying a living room 100% autonomously at real 1x speed with no teleoperation. Your home resets exactly how you like it when you leave.

FOMO? Click here for EVERYTHING that happened in AI today.

FROM OUR PARTNERS

Introducing Tabnine Enterprise Context Engine

Your AI agents are guessing. And it is costing you more than you think.

For the average 500-person engineering organization, the absence of enterprise context leads to $2.4M in AI code rework, code review cycles that run 3.2x longer, and 63% of AI-generated merges failing and requiring rework..

Get a Demo

Tuesday Tool Tip

A Cat’s Commentary

P.P.S: Love the newsletter, but only want to get it once per week? Don’t unsubscribe—update your preferences here.