Recent workPilot

Turning 20 years of work into searchable intelligence

One governed search across terabytes of archives, and one consolidated company database.

Visual for turning 20 years of work into searchable intelligenceA global research and advisory firm
Problem statement

What the client was up against

Fifteen to twenty years of project work sat across a 6 to 8 TB archive, Microsoft 365 backups, SharePoint and huge PST files. Finding an answer meant knowing which folder held it, and company intelligence was trapped in project spreadsheets.

Our approach

How we approached it

  1. Indexed approved sources without moving them
  2. Added OCR, email chain processing and version grouping
  3. Converted deliverables into governed company records
  4. Published changes only after human review, with source evidence
Solution we created

What we built

EXTRCT archive intelligence with one search experience across approved sources and one consolidated company database with deduplication, recency rules and conflict flags.

Agents in the system

The multi agent team behind it

SourcesSharePoint, OneDrive, Microsoft 365 backups, PST archives
Search agentFull text, metadata and semantic retrieval
OCR agentBrings scans and archives into search
Entity agentNormalises and deduplicates company records
Conflict agentSurfaces clashing values for review
People decideReview, approve and act
Search agentFull text, metadata and semantic retrieval
OCR agentBrings scans and archives into search
Entity agentNormalises and deduplicates company records
Conflict agentSurfaces clashing values for review
SharePointOneDriveMicrosoft 365 backupsPST archives
What it was built to change

The outcome it is designed for

Less time to find prior work, faster response to research requests and measurable reuse of institutional knowledge.

This is a pilot without published results yet. We show the problem and the build, not numbers we have not measured.

Client names, detailed case studies, demos and references can be shared on request.

Looking for a similar solution?

Tell us your version of this problem. We will walk you through what we built and what a first pilot would look like for you.

Let's talk