Block's Berd: The Trojan Horse of Agentic Commerce

We didn't ask for another AI agent management tool. But Block—the fintech giant behind Square and Cash App—just gave us one, and it's open source. Or is it? The announcement that Block open-sourced Berd, a desktop app for managing AI agents, was met with a mix of curiosity and skepticism. The crypto media called it 'controlled open-source,' a phrase that carries the stench of open-washing. But beneath the surface, Block's move is not just about code—it's a strategic play for the future of agentic commerce, where agents don't just think, they pay. Context: Berd is a desktop application that allows users to manage, orchestrate, and monitor AI agents. Unlike the dominant cloud-based tools like LangSmith, Dify, or Flowise, Berd runs locally—on your machine. This is a deliberate choice, echoing the local-first philosophy that values privacy and user sovereignty. Block, led by Jack Dorsey, has long championed Bitcoin and decentralization. But Berd is not just another tool; it's a bridge between the world of AI agents and the financial infrastructure Block controls. The original report on Berd was thin on details—no license type, no feature list, no technical specs. But that's exactly why we need to read between the lines. Core: Berd's open-source strategy is a masterclass in balancing openness with commercial intent. From my experience auditing DeFi projects during the 2017 ICO frenzy, I've seen how 'open source' can be a hollow promise. We didn't learn from the 2017 ICO boom that transparency isn't just about code; it's about power dynamics. Block's controlled open source—likely a source-available license that restricts commercial use—is not a bug; it's a feature. It allows Block to benefit from community contributions (bug fixes, feature requests, ecosystem growth) while retaining control over the monetization layer: payment integration. We didn't build the internet on the backs of open protocols only to let walled gardens flourish again. Berd's real value isn't the tool itself—it's the gateway to Block's payment APIs. Think about it: if an AI agent can manage tasks on your desktop, it can also make payments. Berd is the front door to a future where agents execute transactions—booking flights, paying subscriptions, even settling micro-transactions. Block, with its Square payment processing and Cash App wallets, is uniquely positioned to be the rails for these machine-driven payments. This is the 'agentic commerce' thesis: the next trillion-dollar market is not human-to-human commerce but agent-to-agent commerce. From a technical perspective, the desktop form factor is a competitive moat. Cloud-based tools require server infrastructure, API keys, and constant internet connectivity. Berd, by running locally, offers offline capability, full data sovereignty, and the ability to interact with the user's file system and local tools. This is a sandbox for agents—a safe environment where they can experiment without cloud costs. But the real innovation is what's missing from the headlines: Berd likely supports multiple model providers (OpenAI, Anthropic, and local models via Ollama), and it's almost certainly designed to integrate with Block's payment SDK. I've seen this pattern before—when a company opens a tool to build an ecosystem, the real asset is the network effect. Let's talk about the competition. The AI agent management space is crowded: LangSmith, Dify, CrewAI, n8n, and the new OpenAI Agents SDK. Most are cloud-based, some are open source. But none have a built-in payment infrastructure. Berd's unique selling point is that it can eventually enable agents to pay—and that's a game-changer. We didn't see this coming from a payment company, but it makes perfect sense. Block is not trying to beat LangChain at developer tools; it's trying to own the transaction layer of the agent economy. The risk? If the license is too restrictive, the community may reject it. Elastic and Redis faced backlash for similar moves. But if Block gets it right—offering a generous open-source core and a clear path to commercial features—it could become the default agent terminal for developers who value privacy and want to build agentic commerce applications. Contrarian: The narrative of 'controlled open-source' as a betrayal of the open-source ethos is too simplistic. The reality is that sustainable open source requires a business model. We didn't learn from the 2020 DeFi boom that protocols without revenue streams die. Block is being pragmatic: it opens the code to foster adoption and trust, but it retains the right to monetize the most valuable part—the payment integration. This is no different from Red Hat's model, or even GitLab's. The alternative is to close the source entirely, which would be worse for the community. Berd's controlled open source is a compromise, not a crime. It's a recognition that to build a decentralized agent economy, we need infrastructure that is both open and economically viable. Takeaway: The question isn't whether Berd is truly open source. It's whether we, as a community, will build the decentralized agent economy that Block's infrastructure could enable, or let it become another walled garden. The choice is ours. We didn't come this far to hand over the keys to the agent economy to a single corporation—even one with Jack Dorsey's Bitcoin ideals. But if we can steer Berd toward interoperability, open standards (like A2A and MCP), and true decentralization, this desktop app could be the seed of a new financial system where agents transact freely, without intermediaries. The future of agentic commerce is being written—and it's running on a local machine near you.