InoGen

AI Strategy for Remediation Projects

Evidence Discovery and Cost Recovery for a National Housebuilder
Construction
AI Strategy
Legal
£25M value delivered

£20-30M

potential recovery uplift estimated

Keyword search found to miss 40-60% of relevant documents

Full document landscape mapped for the first time

Governance sign-off secured from Legal and InfoSec before implementation

Quantified a £20-30M recovery uplift opportunity for a national housebuilder's fire remediation programme by assessing the gap between keyword search and AI-assisted document discovery. Delivered a validated tool recommendation, information governance model, and phased rollout plan that converted an ambiguous AI discussion into a funded, board-approved delivery programme.

The Problem

Following the Grenfell Tower tragedy, national housebuilders faced large-scale fire remediation programmes. Recovering costs from liable parties requires evidence: contracts, specifications, correspondence, and inspection records stretching back years or decades. This housebuilder's evidence sat across a vast landscape of digital repositories, legacy file shares, archived email systems, local drives, and physical filing cabinets in regional offices. No single index existed. The volume ran to hundreds of thousands of files across hundreds of developments.

The Solution

The engagement began not with a tool selection but with a structured assessment of whether AI-assisted document discovery could credibly unlock enough additional recovery value to justify the investment. We ran a three-phase strategy engagement.

Phase 1: Value Validation. We catalogued the full set of document sources across the estate, documenting volumes, format mix, indexing status, and access models. The picture was worse than expected: several substantial repositories had never been searched because no one on the current team knew they existed. We then assessed the evidence gap by taking a sample of developments where keyword search had been run and applying more thorough review (supplemented by pilot OCR and semantic search). Keyword search was surfacing roughly 40 to 60 percent of relevant documents. Extrapolating across the full programme gave us the benefit estimate: a potential uplift of £20 to 30 million in recoverable costs.

Phase 2: Solution Assessment. We built a scoring framework across five weighted dimensions (coverage, search quality, extraction accuracy, security, and workflow fit) and evaluated three off-the-shelf platforms and two bespoke-build options. Shortlisted candidates were tested in controlled proof-of-concept exercises against documents where ground truth was known, exposing practical limitations that demonstrations alone would have concealed.

Phase 3: Rollout Planning. The information governance model was defined early in collaboration with legal, compliance, and information security teams, covering data residency, encryption, access control, retention, and audit logging. Engaging these stakeholders at the strategy stage rather than the implementation stage meant the recommendation arrived at the programme board with their concerns already addressed. The rollout was sequenced to align with programme priorities: high-value developments with the largest evidence gaps went first, generating early recoveries that demonstrated value and built confidence.

Loading diagram...

Results and Impact

MetricOutcome
Potential recovery upliftEstimated at £20 to £30 million across the remediation programme
Evidence gap quantifiedKeyword search was missing 40 to 60 percent of relevant documents
Document landscape mappedFull taxonomy of sources, formats, volumes, and access models produced for the first time
Solution recommendedSpecific tool selected and scored against five-dimension evaluation framework with PoC validation
Governance secured earlyInformation governance, security, and retention model agreed with Legal and InfoSec before implementation
Delivery plan definedPhased rollout aligned to remediation programme priorities, with operating model and training plan

The strategic assessment converted an ambiguous "we should probably use AI for this" conversation into a concrete, costed plan with stakeholder buy-in. The housebuilder moved from knowing they were missing evidence to having a quantified gap, a validated tool recommendation, and a delivery path the programme board could approve with confidence.

Key Takeaways

  • Quantifying the evidence gap made the business case undeniable. Showing that keyword search was missing 40 to 60 percent of relevant documents turned the conversation from "should we do this?" to "how fast can we start?"

  • Early governance engagement avoided later blocks. Legal, compliance, and information security teams have the authority to stop technology deployments. Co-designing the governance model at the strategy stage saved months compared to building first and seeking approval later.

  • Scoring vendors against real workflows stopped shiny-tool bias. By evaluating against coverage, search quality, extraction accuracy, security, and workflow fit with real sample data, the assessment surfaced practical limitations that demonstrations alone would have concealed.