Before
Manual and fragmented
Project knowledge and templates live across a large SharePoint library. Finding the right source, reading it and turning it into a useful draft interrupts delivery teams and produces inconsistent starting documents.
Quartz Project Services
Leo AI answers project questions and generates working documents inside Microsoft Teams by searching the construction firm's own SharePoint files.
How the work moves
Client
Quartz Project Services
Agenryx system
AI & Automation
Operating change
Staff ask in Teams and receive a grounded answer in the same thread
Live
Client system operating in Microsoft Teams
2
Independently deployable services
1
Grounded SharePoint search tool
01 · The challenge
What the client needed to resolve
Project knowledge and templates live across a large SharePoint library. Finding the right source, reading it and turning it into a useful draft interrupts delivery teams and produces inconsistent starting documents.
The change
Before
Project knowledge and templates live across a large SharePoint library. Finding the right source, reading it and turning it into a useful draft interrupts delivery teams and produces inconsistent starting documents.
Agenryx built
A two-service Railway deployment: a Node and TypeScript Teams bot plus a Python and FastAPI MCP server, each containerised independently and joined by a local Compose setup for development and diagnosis.
Resolved
Staff ask in Teams and receive a grounded answer in the same thread
02 · What we built
A Teams webhook calls a separate MCP service. That service searches and fetches SharePoint content through Microsoft Graph, gives the retrieved material to a headless document agent and returns the answer or generated file links to the originating thread.
A two-service Railway deployment: a Node and TypeScript Teams bot plus a Python and FastAPI MCP server, each containerised independently and joined by a local Compose setup for development and diagnosis.
03 · Product walkthrough
Select a step or let the walkthrough advance automatically.
Project knowledge and templates live across a large SharePoint library. Finding the right source, reading it and turning it into a useful draft interrupts delivery teams and produces inconsistent starting documents.
04 · What changed
Staff ask in Teams and receive a grounded answer in the same thread
Document requests reuse the firm's own SharePoint templates
Generated files return through signed download links instead of email attachments
Technology used