Product update

FlowerDocs configuration and extension is now AI-driven

July 29, 2026 · Product · 3 min read

Solène Boheas

Product Manager · Uxopian Software

What if you could divide by five the time it takes to deliver your FlowerDocs backlog? This is now a reality: the FlowerDocs MCP Server puts an AI assistant to work on the configuration of the platform itself.

AI's blind spot in document management

Over the past two years, every document management vendor has added an AI layer. Automated classification, data extraction, document chatbots: AI helps users get more out of what's already in the system.

But before a user can search, classify or automate anything, someone has to configure the platform. Document classes, workflows, search forms, access rights: it's deep, technical work that requires understanding how all the pieces fit together.

And yet, that's where project success is decided.

FlowerDocs opens its configuration to AI

FlowerDocs is Uxopian Software's document management platform, deployed across insurers, mutual insurance companies and organizations managing high volumes of business documents.

With Uxopian AI, we already help users get more from their documents. With the FlowerDocs MCP Server, we've chosen to go further: applying AI to the setup of the platform itself. The MCP server connects an AI assistant directly to the administration layer.

The MCP server equips an AI assistant with hands-on access to the FlowerDocs administration layer. The administrator describes what they want in natural language: create a class, adjust a workflow, configure a search form ... The assistant turns the request into actions, drawing on its knowledge of how FlowerDocs objects depend on one another to guide and safeguard each step. What used to require dedicated training and deep platform knowledge is now accessible from day one.

science

Field test: better managing RFPs with FlowerDocs

Fresh FlowerDocs tenant · MCP server installed · Claude Code as the assistant

2 min Declare the MCP server in your Claude MCP client.

10 min Iterate on the specifications with the assistant, which reads the FlowerDocs documentation exposed over MCP, and draft a first spec.

30 min Review the plan, adjust a few points, and let it build.

Half an hour later, the module was there: document types configured with the right metadata, virtual views that make everyday work easier for business users, and roles defined and wired.

It even shipped with the system to manage and approve a bank of validated answers, and an advanced handler that splits the incoming Excel sheet, reuses approved answers, and opens workflow tasks for the questions with low confidence or no answer at all.

Three document types · a few handlers · a workflow class · user actions, teams, roles, custom ACLs

A working RFP module in under one hour

Don't get us wrong: this is where the real work starts. Our recent vibe-coding experience taught us that it can take many iterations to find the right spot for your business users. Those iterations were always necessary; in the past, there was simply no budget left to run them. Now there is. This is the time to make your projects shine.

lock

Every action runs under the permissions of the connected FlowerDocs user. The assistant can never do more than that user is already allowed to do.

What this changes for a FlowerDocs project

1

Faster onboarding

Configure FlowerDocs without prior training.

expand_more

The real complexity of FlowerDocs configuration was never the product itself, it was knowing how objects relate to one another. The assistant handles that complexity. An integrator can start a project without prior training, even on technical configurations.

2

Always best-practice compliant

The assistant follows current guidelines, not old habits.

expand_more

Paired with the FlowerDocs documentation, the assistant produces configuration that follows current best practices. Quality stays consistent regardless of the integrator's experience level.

3

Ask your configuration directly

Query what's in place instead of rebuilding it by hand.

expand_more

Which document types use this metadata? Which workflows will be affected if I change this configuration? The assistant answers without navigating the admin interface. Whether it's for incident analysis, impact assessment or bringing project documentation up to date.