A healthcare system wanted to roll out an internal LLM-based knowledge assistant for clinical staff, but its underlying policy and procedure documentation was scattered across six departmental data lakes, most with no consistent tagging or version control.
Consolidating Fragmented Data Lakes Ahead of an Enterprise LLM Rollout
Service Provided
Capabilities Utilized
Data Architecture, Data Quality, Enterprise Search Infrastructure
Client challenge
Clinical staff routinely found conflicting versions of the same procedure document across different departmental repositories, with no reliable way to know which was current. Building a knowledge assistant on top of this fragmented, inconsistently maintained documentation risked confidently surfacing outdated or contradictory guidance — an unacceptable risk in a clinical setting.
Our solution
Before touching the LLM layer at all, we consolidated the six data lakes into a single governed document repository with mandatory version control and department-level ownership for keeping content current. We built automated staleness detection that flags any document not reviewed within its required cadence, rather than assuming documents stay accurate indefinitely.
Only once the underlying repository was clean did we build the retrieval layer for the knowledge assistant, deliberately restricting it to pull exclusively from the governed repository rather than the legacy data lakes, which remained accessible for historical reference only.
Results
The consolidation project eliminated 340 duplicate or conflicting document versions before the knowledge assistant ever went live, and the staleness-detection system has since flagged an average of 12 documents per month for mandatory review. Clinical staff adoption of the resulting knowledge assistant reached 76% within three months, with a measured document-retrieval accuracy of 97% against a validation set reviewed by clinical informatics staff.
Additional Case Studies
Launching a Usage-Based Pricing Model for an AI Developer Platform
An AI developer platform’s flat-tier pricing was leaving money on the table with its largest customers
Establishing an Enterprise AI PMO to Track ROI Across 30 Initiatives
A global enterprise had 30 active AI initiatives across its business units but no consistent way
Building the Product Operating Model for an Agentic SaaS Vendor
A vertical SaaS vendor had shipped an early agentic feature that could autonomously complete customer workflows,
