Block Launches Berd to Replace AI Chatbots With Agentic Canvas
The new open-source workspace moves away from traditional chat windows toward a multi-model 'crew' of agents.
Block has launched Berd, an AI-centric workspace designed to shift the user experience from traditional chatbot interfaces toward a customizable canvas. The platform aims to reduce the friction of AI interaction by prioritizing a building-centric environment over standard conversational windows.
Developed by Block (block.xyz), Berd is an open-source project built upon the "goose" AI agent framework. Rather than interacting with a single prompt box, users can construct a "crew" of agents, each equipped with distinct personalities and specialized skills. The workspace allows for deep integration with developer workflows, enabling users to link their projects directly to existing folders and repositories. To optimize for both cost and computational power, the platform supports multi-model flexibility, allowing users to switch between different LLM providers depending on the complexity of the task at hand.
The Shift to Agentic Workflows
This launch arrives during a pivotal transition in the AI industry. While the first wave of generative AI was defined by simple chat interfaces, the market is now moving toward "agentic" workflows. These specialized workspaces integrate more deeply with professional tools, treating AI not as a conversational partner but as a set of autonomous agents capable of managing files and executing complex project structures. By moving the interaction to a canvas, Berd attempts to mirror the way developers actually work—visually organizing components rather than scrolling through a linear chat history.
Challenging the Chatbot UX
By positioning the platform as "less chatting, more building," Block is attempting to solve the inherent limitations of single-model dependency and the constant tool-switching that plagues current AI workflows. The "build a crew, not a chatbot" philosophy suggests a future where AI is a modular workforce rather than a singular oracle. Leveraging Block's resources, Berd challenges the dominant UX established by early LLM leaders, betting that professional users prefer a spatial environment where agents can be orchestrated across a project's lifecycle.
What to Watch
As an open-source project, Berd's growth will likely depend on community adoption of the underlying goose framework. While the core functionality is now live, the industry will be watching to see if the canvas-based approach significantly increases productivity over traditional chat. Additionally, the extent to which the multi-model switching simplifies the cost-benefit analysis for enterprise users remains a key point of interest for the developer community.