Safely manage your Zendesk from the AI assistant you already use, via the Deltastring MCP. Beacon configuration platform
← Back to news

Block’s new Apache 2.0 agent workspace Berd works across models and harnesses, stores conversation history locally

Block has open-sourced Berd, a desktop application originally built for internal use that provides a unified workspace for managing AI agents across multiple models. The move reflects a broader industry shift toward interoperability in agent tooling, particularly as enterprises grapple with vendor lock-in concerns and the operational complexity of managing conversations across disparate AI systems. By storing conversation history locally rather than in proprietary cloud infrastructure, Berd addresses a critical pain point for CX teams operating under data governance constraints—a consideration that becomes more pressing as companies race to cut oversight functions after AI failures. The open-source Apache 2.0 licensing signals Block's confidence in the tool's utility whilst removing friction for adoption across different organisational contexts.

For CX professionals, the implications are twofold. First, Berd's model-agnostic architecture challenges the assumption that agent workspaces must be tightly coupled to a single LLM provider or platform vendor. This matters particularly for teams already invested in Zendesk, Salesforce, or similar platforms—the question becomes whether these incumbents will prioritise interoperability or continue betting on proprietary ecosystems. Second, the local conversation storage model offers a technical pathway for teams concerned about data residency and audit trails, though it introduces new operational considerations around backup, synchronisation, and compliance across distributed deployments. The release suggests that the competitive advantage in agent tooling is shifting from closed platforms toward flexibility and transparency, which could accelerate consolidation among vendors unable to offer genuine multi-model support.