While the market fixates on the next Fed pivot and the liquidity flush from rate cuts, a quieter signal emerges from Jack Dorsey’s Block: the open sourcing of Berd, a desktop AI agent manager. This is not a developer tool. It’s a plumbing play for the next wave of automated commerce. And the crypto community, via Crypto Briefing, is the first to notice. But they’re reading it wrong. They see a limited open-source release. I see a macro shift: the convergence of local-first infrastructure, payment rails, and agentic autonomy. The plumbing is being laid, not for the next meme coin, but for the machine-to-machine economy. Don’t watch the price; watch the plumbing.
The Context: What Is Berd and Why Should You Care?
Berd is a desktop application for managing AI agents—think of it as a local-first IDE for orchestrating language model workflows. Block, the company behind Square and Cash App, open-sourced it under what the article calls a “controlled open-source strategy.” No license details, no feature list, no technical specifications. Just a code drop and a promise of accessibility. The analysis I reviewed, from a deep-dive report, classified Berd as a “combinatorial innovation”—a wrapper around existing GUI frameworks, agent orchestration, and local-first paradigms. Not a breakthrough. But the strategic intent is clear: Block is using Berd as a wedge into the AI agent infrastructure layer.
From my perspective as a digital asset fund manager who has watched the ebb and flow of tokenized economies since 2017, this is a familiar pattern. The same structural logic that drove the 2017 ICO audits—technical integrity precedes market value—applies here. Berd’s desktop form factor is a deliberate choice. In a world where every other agent management tool (LangSmith, Dify, Flowise) is a web SaaS, Block is betting on local-first. Why? Because local-first is the only way to give agents safe access to system resources: file systems, local APIs, and—critically—payment cards. The plumbing is local, the incentives are central.
But the article’s core claim is that this is “controlled open source” that limits community collaboration. I’d argue that’s a misreading. The control is the point. It’s a strategic moat, not a limitation. Let me unpack that.
Core Insight: The Real Value Is in the Payment Rails, Not the Code
Let’s start with first principles. Block’s primary revenue comes from transaction fees—every swipe of a Square terminal, every Cash App transfer. Its business model is built on capturing value from the flow of money. Now, consider the next evolution: AI agents that can act autonomously, booking travel, subscribing to services, or purchasing inventory. For an agent to complete a real-world transaction, it needs a payment channel. The agent management tool is the gateway to that channel.
Berd’s open-source code is bait. The real product is the integration with Block’s payment API. This is not speculation; it’s pattern recognition. In 2025, Stripe launched its own agent toolkit, explicitly targeting the same “agentic commerce” use case. Block is a year behind, but it has two advantages: (1) a massive installed base of merchants and consumers via Square and Cash App, and (2) a Bitcoin-native infrastructure through Spiral and TBD. Jack Dorsey’s obsession with decentralized identity and Bitcoin means Berd could eventually support Lightning Network payments or Web5 identity verifications. That would make it a crypto-native agent tool, not just a fintech play.
But the article’s technical analysis stops at the surface. It correctly notes that Berd likely supports multiple model providers (OpenAI, Anthropic, local Ollama)—a standard feature. The hidden information, however, is the sandboxing. A desktop agent manager can run agents in a sandboxed environment, limiting their access to the host system. That’s a security feature that cloud-based tools struggle with. And if Block integrates its fraud detection and AML/KYC frameworks into Berd’s agent authorization layer, it becomes a compliance-ready platform for regulated industries. That’s the moat. Code is law, but incentives are god.
I’ve seen this before. In 2020, during DeFi Summer, I witnessed yield farming protocols that promised 1000% APY but were built on debt ponzis. The plumbing was flawed. The sustainable model was the one that controlled the liquidity flow—like Aave or Compound. Berd is Block’s attempt to control the flow of agent-driven transactions. The yield is not in the tool; it’s in the transaction volume it generates.
Contrarian Angle: The Decoupling Thesis Applied to AI Infrastructure
Now, the contrarian perspective. The market narrative around AI agents is that they will decouple from traditional infrastructure—a kind of technological decoupling similar to what crypto believers hoped for in 2020: that crypto would decouple from traditional markets. That didn’t happen. Crypto correlated with risk assets. But the agent infrastructure decoupling is real. Large language models are becoming commoditized; the value is shifting to the orchestration and payment layers.
Berd is a bet on this decoupling. By being desktop-first, it decouples from the cloud. By being open-source (even if controlled), it decouples from proprietary vendor lock-in. By being built by a financial company, it decouples from the pure AI narrative and anchors itself in the real economy. The article’s analysis of competition is accurate: Berd is early, its ecosystem is weak, and the developer community is skeptical of Block as an AI company. But the decoupling thesis suggests that the winner in this space will not be the best developer tool but the one that controls the most valuable transaction flows.
Consider the 2022 Terra collapse. Everyone blamed the algorithmic stablecoin design. My thesis was that the real cause was excessive dollar-denominated leverage—a macro liquidity shock. The plumbing was broken. Berd’s plumbing is sound because it’s local-first. The risk is not technical failure but market adoption. If Block fails to attract developers, the project dies. But if it succeeds, it could become the default agent manager for Bitcoin-native agents, creating a new asset class: machine-generated transaction fees.
Bubbles don’t burst because of fraud; they burst because of liquidity. The liquidity in agent economics is the trust that agents can execute payments without human intervention. Berd is a vote of confidence in that future.
Takeaway: Positioning for the Next Cycle
Berd is not a product. It’s a strategic option. The forward-looking judgment is this: watch the integration of Berd with Block’s payment API. If, within the next 12 months, developers can build agents that use Square SDK to transact, the market will price Block’s stock accordingly. The crypto community should watch for Lightning Network support. If Berd becomes the first desktop agent manager with native Bitcoin payments, it will be a catalyst for the next wave of Bitcoin adoption—not as a store of value, but as a medium of exchange for machines.
My own experience from the 2024 ETF institutional pivot taught me that the slow, steady integration of blockchain into traditional balance sheets is where the real value lies. Berd is that integration for AI agents. The question is not whether this will happen, but who will control the plumbing. Block is betting that local-first, payment-integrated, and crypto-native is the right stack. I’m betting with them.