Block’s Berd Open-Source: The Trojan Horse for Agentic Commerce?

The story isn’t in the token, it’s in the trust. Last week, Block (Jack Dorsey’s fintech empire) quietly open-sourced Berd, a desktop app for managing AI agents. The announcement from Crypto Briefing was sparse—no license details, no feature list, no technical specs. But the silence itself is a signal. When a company that processes billions in payments releases a tool that lets agents orchestrate actions, the real story isn’t the code. It’s the pipeline. This isn’t just another AI agent dashboard. LangSmith, Dify, Flowise—they all live in the cloud. Berd lives on your desktop. That choice is a philosophical statement: local-first, privacy-first, and—if you read between the lines—Bitcoin-first. Jack Dorsey has been preaching “open AI + Bitcoin” for years. Berd is the first concrete product of that vision. Let’s rewind. The AI agent management space is crowded. LangChain’s ecosystem, OpenAI’s SDK, Microsoft’s Copilot Studio—they’re all fighting for developers’ mindshare. But none of them own the payment rails. Block does. Square processes millions of merchant transactions. Cash App moves billions in peer-to-peer payments. And now, with Berd, Block is building the operating system for agents that can spend money. From my time moderating the Ampleforth Discord in 2020, I learned that the most valuable infrastructure isn’t the one with the most features—it’s the one that reduces friction for the end user. Berd’s desktop form factor eliminates the need for cloud deployment. A developer can download it, connect a local LLM via Ollama, and start building an agent that books flights, orders groceries, or pays invoices. The agent stays local. The payment flows through Block. That’s the loop. The core insight here is about trust architecture. In the 2021 meme economy, I saw how narratives preceded utility. The same is happening now with AI agents. The technology is raw, but the narrative is already forming: agents will do real-world tasks, and they will need to pay. The company that controls the agent’s wallet controls the future of commerce. Block is betting that by giving away the agent management tool for free (under a controlled open-source license), they can capture the payment flow. It’s a classic razor-and-blades model: give away the razor, sell the blades. But the “controlled” part is key. The Crypto Briefing article framed it as a limitation—a restriction on community collaboration. I see it differently. From my work with institutional clients in 2024, I learned that trust is not binary. It’s layered. A fully open license might let anyone fork the code and build a competing payment rail. A fully closed license would kill developer adoption. The middle ground—source-available with restrictions on commercial use—is the sweet spot. It allows Block to maintain the link between the agent and its payment infrastructure while still getting the community’s bug fixes, security audits, and ecosystem extensions. Let me be contrarian for a moment. The narrative says “open source is always better.” But in the AI agent space, the real value isn’t the code—it’s the network of trust that connects the agent to the world. An agent that can’t pay is a toy. An agent that can pay but uses a random API is a security risk. Block’s moat is not its code; it’s its compliance infrastructure, its fraud detection, its KYC/AML systems. That’s what makes Berd dangerous to competitors. Stripe launched an agent toolkit in June 2025. But Stripe doesn’t have a desktop app. It doesn’t have a local-first philosophy. It doesn’t have Jack Dorsey’s Bitcoin maximalist community. There’s a blind spot here, though. The crypto community might reject Berd precisely because it’s centralized. Block is a corporation. The payment rails are still fiat. Jack Dorsey’s Bitcoin dream is real, but Berd today doesn’t connect to Lightning Network or any on-chain payments. If Block moves too slowly, a decentralized alternative—like a DAO-operated agent marketplace with crypto payments—could eat their lunch. The winter of 2022 taught me that communities survive by holding hands. The bear market bonded the survivors. The same will happen in the agent space: the open, permissionless, trust-minimized version will win if the centralized version falters. But let’s look at the data. The sentiment around Berd on crypto Twitter is mixed. Some see it as a sign of mainstream adoption. Others see it as a co-opting of the AI agent narrative by traditional finance. The on-chain volume of agent-related tokens spiked on the announcement, but that’s noise. The real signal is the developer activity on Berd’s GitHub repo. I haven’t seen the code yet, but based on the product description, I expect to see: (1) multi-model support including local LLMs, (2) a plugin system for tools, and (3) an API for payment integration. If the API is there, then Block is serious. If not, it’s just a toy. From my experience building the “Empathy Algorithm” for AI-DAO hybrids, I know that the most successful agent systems are those that combine technical efficiency with human-centered narrative. Berd’s desktop form factor is a nod to that: it puts the user in control of the agent’s data and actions. The agent operates in a sandbox on your machine. You can see every tool call, every API request. That transparency is crucial for trust. In the 2026 landscape, where AI agents are autonomous but still need human oversight, Berd’s local-first design could be a competitive advantage. So what’s the takeaway? The next narrative in crypto is not “AI agents” generically. It’s “Agentic Commerce”—the idea that agents will act as autonomous economic actors, earning and spending money. Block’s Berd is the first major attempt by a traditional payments company to plant a flag in that space. The contrarian view is that this is a Trojan horse: a centralized payment rail disguised as an open-source tool. But maybe that’s what we need to bridge the gap between the crypto world and the real economy. The story isn’t in the token. It’s in the trust. And trust is built one local, transparent, permissioned agent at a time. The question remains: will Berd become the standard for agent payments, or will a decentralized alternative emerge from the crypto community? The answer depends on whether Block opens up the payment rails to Bitcoin and Web5. If they do, they’ll have the community. If they don’t, they’ll have a fork. And in the winter of 2022, we learned that the community always wins. — A. Chen, Vienna, 2026