When Reports Talk to Each Other: Elevent Gives a Voice to Its Institutional Memory
How an event management company built an AI agent directly into Claude to speed up report production and facilitate knowledge transfer.
Elevent, an event management company, had hundreds of archived reports sitting on a wealth of untapped information. Mondrian built an AI agent integrated directly into Claude Desktop via the MCP protocol, letting teams query the entire archive in natural language from an interface they already used every day. Rather than building a custom tool from scratch, Mondrian connected Claude's capabilities directly to Elevent's knowledge base — so adoption was immediate and use felt natural. The solution now serves 12 daily active users and saves 1 to 2 hours per report produced.
An organization whose real wealth sits in its archives
Event management and production
Québec, Canada
Elevent is an event management company that produces, season after season, a large volume of documented deliverables: performance reports, post-event analyses, and strategic summaries. Over time, this output generates a considerable knowledge base, but one that is difficult to leverage on a day-to-day basis. Recognizing that its historical data represents an underutilized asset, Elevent decided to invest in an AI solution to make this knowledge accessible and actionable for both its experienced teams and new employees.
Hundreds of reports on hand, but knowledge that stayed out of reach
Elevent's teams regularly produce reports that draw on past data and analysis. Before the solution, tracking down the right report among hundreds of archives meant a slow, unreliable manual search. The information existed — but finding and using it depended entirely on individual memory and experience.
This friction slowed down the production of each new report by 1 to 2 hours. It also created a risk of knowledge loss during team transitions: new employees had to rely on their colleagues to navigate the archives, placing an additional burden on teams that were already stretched thin.
Three constraints made the situation complex. First, the 342 archived documents were not structured uniformly: they had heterogeneous metadata, were in a mix of languages, and came in various formats. Second, a traditional search solution would have been insufficient: the teams needed to understand the content of the documents, not just find them by keyword. And third, any new interface risked going unused if it added friction to already-busy workflows.
An agent integrated into Claude: Elevent's expertise, right where teams are already working.
An MCP server deployed on Google Cloud Platform exposes document search capabilities directly inside Claude's interface. When a user asks a question, Claude queries Elevent's 342 documents — indexed in a Weaviate vector database — behind the scenes, contextualizes the results, and presents them in a directly usable form. The tool supports both report production research and new-hire onboarding, letting new employees freely query the organization's institutional memory from day one. Access is secured through WorkOS AuthKit, ensuring only authorized team members can interact with the organization's data.
Mondrian worked closely with Elevent's teams to structure and standardize document metadata before ingestion — a critical step for search result quality. Users were brought in early through a testing phase that helped refine the solution before full deployment. Choosing to build on Claude Desktop rather than a custom tool was the key decision that ensured fast, friction-free adoption.
| Component | Role in the solution |
|---|---|
| Claude Desktop + MCP | Primary interface: the agent is accessible directly in Claude, with no new tool to learn and no training required |
| MCP Server on GCP | Secure bridge between Claude and Elevent's knowledge base, deployed and maintained on Google Cloud |
| Weaviate (vector database) | Indexing and semantic search across 342 archived documents |
| WorkOS Authkit | Secure authentication, limiting access to authorized users only |
Immediate adoption, measurable value from the very first use
| Indicator | Result | Business impact |
|---|---|---|
| Documents indexed | 342 archived reports, fully searchable | Elevent's entire institutional memory accessible in seconds, right from Claude |
| Time saved per report | 1 to 2 hours per deliverable | Direct productivity gains for teams in every production cycle |
| Adoption | 12 active users, daily queries | Adopted naturally thanks to integration into a tool teams already use |
| Knowledge transfer | Go-to reference tool for new hires | Less reliance on experienced colleagues during onboarding |
Facing a similar challenge in your industry?
A 30-minute call is enough to find out where AI can create value for you.
Book a call