E-Discovery Platform for Forensic Document Review

The Challenge

External e-discovery providers charge up to €25 per gigabyte for processing and €15 per gigabyte for hosting. For a typical case volume of 100 GB, that adds up to several thousand euros paid to an intermediary - before a single document has been reviewed.

For large proceedings, this pays off. For small and mid-sized cases in the 8 to 25 gigabyte range, it does not. This is precisely the segment established platforms do not serve economically. Our client, a newly founded platform operator working alongside a law firm, set out to close that gap with a platform of their own.

The requirements were strict. Client data falls under legal professional privilege. Every access must be traceable, matters must remain cleanly separated, and operations must take place entirely within the EU.

Our Approach

We took responsibility as lead architect, from target architecture through implementation - structured across five milestones with a fixed go-live date.

Event-driven ingestion. Data from mailboxes, chats and file shares flows through a queue into scalable workers. Each source has its own parser - PST, Office documents, Teams, Slack and further formats. Scanned documents run through local OCR, so no data leaves the server.

One database instead of four. Rather than operating separate SQL, NoSQL, full-text and vector stores, PostgreSQL with JSONB and pgvector covers the majority. OpenSearch handles full-text search - deliberately chosen because its licence is safe for the platform model. Original files live in S3-compatible object storage.

Matter separation and auditability. Logical separation at the database layer, a complete audit trail across all review steps, and a security concept built to withstand the external audit ahead of go-live.

On hosting we deliberately made no advance decision. Instead we produced a decision template scored against nine KPIs - EU hosting, data processing agreement, backup recovery time, cost structure at 100 GB per case. The client made the call.

The Results

The platform replaces the intermediary provider entirely. Cases that previously could not be taken on for cost reasons are now economically viable.

Just as important as the platform itself is the operational documentation. Architecture, deployment and CI/CD are documented such that a competent third party could rebuild the system. The client is meant to operate the platform themselves long term - that was part of the objective from the outset, not an afterthought.