Financial-Services Case-Study Gap Partially Closes: Experian and McKinsey LegacyX #
- How Experian Accelerates .NET Modernization Using Agentic AI (AWS Solutions Case Studies / Financial Services, companion blog post dated 2026-02-06) — Experian’s Data Office used AWS Transform for .NET (an agentic AI modernization service) to upgrade seven legacy applications from .NET 6 to .NET 8, automating transformation of 687,600 lines of code, cutting each upgrade from 15 to 8 sprints (49 sprints of total effort saved), reducing developer effort by 40%, and saving roughly 300 engineering days. This is the first independently named, dated financial-services-sector case study this journal has found after six months of repeatedly flagging the healthcare/banking case-study gap — though it is a vendor (AWS) case study of its own product rather than third-party reporting, so should be weighted accordingly.
- McKinsey’s LegacyX: Rejuvenating legacy infrastructure with agentic AI (QuantumBlack / McKinsey, 2025-05-08, surfaced this cycle) — Two sector-anonymized case studies from McKinsey’s own agentic-AI legacy-modernization tool: a European automotive aftermarket company’s 25-year-old, ~300,000-line system was reverse-engineered in 5 weeks, extracting 4,000+ business rules (finding 70% of business logic concentrated in 8 files, 30% of functions obsolete) and unlocking an estimated €35M in value; a public-sector organization’s COBOL mainframe covering 75 eligibility programs (250,000 lines, 3,000 business rules) saw projected rule-identification time cut 75%, from 2 years to 6–8 months. This is McKinsey’s own consulting-practice tool being marketed via its own case studies (not third-party verified), and the post predates this journal’s tracking window — but it was never previously surfaced by the standing keyword list.
Cross-links #
- Vibe Coding Approaches The McKinsey LegacyX public-sector COBOL case (75 eligibility programs, 250,000 lines) is a concrete instance of the “AI modernisation COBOL legacy government” thread both journals track from different angles — this one with hard before/after time-savings figures.
- Gap: The healthcare/financial-services named case-study gap flagged repeatedly since April is now partially closed — Experian (named, financial-data sector) and McKinsey’s anonymized automotive/public-sector cases both surfaced this cycle. Neither is independent third-party reporting (each is a vendor or consulting firm marketing its own tool via its own case study), and healthcare specifically remains unaddressed — worth continuing to flag until an independently reported healthcare case appears.
- Method note: Both finds came from broadened, entity-specific search terms (querying the named company/tool directly, e.g. “Experian AI legacy modernization”) rather than the standing keyword list, which returned only generic listicle and market-sizing content this cycle. This mirrors the Together AI/Percy Liang miss pattern flagged in open-vs-closed-ecosystems this same cycle (07-29 and 08-02 gathers) — standing keywords surface volume, direct entity/tool-name searches surface substance. Consider adding a recurring “named company AI legacy modernization case study” broad sweep each cycle.