Lucid Software has expanded its enterprise AI toolkit with updates to its Model Context Protocol server, broader Lucid AI capabilities, and the launch of a new Process Agent designed to help teams build diagrams and documentation faster. The company said the changes are aimed at reducing the gap between an early idea and the structured assets teams need to execute work.
The update builds on Lucid’s MCP server, launched in November 2025, which lets users search, fetch, and summarise documentation from across their technology stack inside large language models. The latest release adds in-LLM diagram creation, so users can describe a workflow and generate a fully editable visual directly within their chosen model interface.
Lucid said further additions are on the way. A ChatGPT Apps SDK integration is listed as coming soon, which would allow users to create editable Lucid diagrams and search existing documents from within ChatGPT. An Edit Document API is also due in April, allowing users to describe changes inside an LLM while Lucid AI carries out the edits automatically.
Alongside the MCP update, the company has widened the reach of Lucid AI across its suite. New capabilities include voice-to-text prompting, swimlane support with colour-coding and nested containers, faster board generation, and layout tools that let teams move between high-level views and more detailed documentation without losing structure on the canvas.
The new Process Agent is positioned as the most significant step in the release. Rather than simply responding to a prompt, it is designed to ask follow-up questions about scope, approvals, and risk before a diagram is produced. Lucid said that should help teams create process documentation that better reflects real business logic and internal compliance requirements, particularly in areas such as change control and infrastructure planning.
Jamie Lyon, Chief Product & Strategy Officer at Lucid Software, said: “For most organisations, the distance between an idea and an enterprise-ready asset that teams can actually execute on is wider than it should be.”
That framing goes to the centre of the company’s pitch. Lucid says only 15% of UK knowledge workers describe their company’s workflows as “extremely well-documented”, leaving teams without the shared context needed to make fast decisions or apply agentic AI effectively. In practice, that means the value of another model upgrade may depend less on the model itself than on whether an organisation can surface and structure its own process knowledge.
Lucid’s answer is to position its platform as a “system of action” rather than another system of record. The idea is straightforward: bring scattered documents, process maps, and decision points onto a shared visual layer that both people and AI agents can use. The company said its MCP server connects with a wide range of clients, including ChatGPT, Claude, Microsoft Copilot, GitHub Copilot, and Cursor.
For enterprise teams working through AI roll-outs, the release is less about novelty than operational clarity. Lucid is betting that the organisations able to document, visualise, and refine work quickly will be better placed to turn AI experimentation into repeatable execution.
