Wipro Health Plan Services — AI Transformation & Modernization
Wipro Health Plan Services is rebuilding its platform off the mainframe and infusing AI across the business at the same time — new cloud-native products built with agentic development and our AI SDLC, and AI used to modernize the legacy estate. This is the program record: the approach, the delivery engine, the platforms already landed, and the decisions now on the table.
01 — The thesis
Most modernization programs ask for years of patience before value appears. This program was structured the opposite way: two tracks, quarterly proof, and an evidence system that makes every claim inspectable.
Platform modernization runs alongside AI transformation. Each track de-risks the other: new cloud-native systems shrink the legacy surface, while the AI delivery engine compresses the time each system takes to build.
Requirements trace from source to backlog to gate to test to artifact. Savings are counted when workloads are verifiably retired — not when code is migrated. External claims wait for completed audit and certification evidence.
The GroupLink Portal coexistence bridge, the MarketLink regulated delivery pathway, and the AI SDLC itself are patterns, not one-offs — each one lowers the cost and risk of the next platform.
02 — The starting position
The program began against a fixed-cost, vendor-operated mainframe estate. The platform worked — but product change moved at legacy speed, cost scaled linearly with membership, and the specialist talent pool kept shrinking.
Baseline: January 2026 strategic business case and the April 10, 2026 mainframe migration assessment kickoff. COBOL, JCL, and DB2 skills concentrated in a diminishing talent pool; managed-services contract runs to 2030.
03 — The delivery engine
Delivery velocity comes from a documented, model-agnostic operating method — Agile discipline adapted for agentic delivery. Agents execute bounded work in scoped workspaces; humans own the gates; every release ships with a typed evidence packet.
Thirteen security and compliance gates sit on top of AI-assisted delivery. Security is a release condition.
The method is grounded in NIST AI RMF, NIST SSDF, and OWASP GenAI guidance, and it is deliberately tool-agnostic: models and tools can change while policy, evidence, security gates, and human authority stay constant. The same governed lane has now delivered regulated, security-reviewed platform work across four product families.
04 — The proof
Each platform was built inside the same governed engine, and each one retired a different objection: too regulated, too legacy, too operational, too entangled.
A CMS EDE broker and agent enrollment platform taken from concept to auditor handoff in six weeks — against an 8–12 month conventional planning baseline — with 20 CMS/EDE build areas complete and a 100% requirement traceability spine.
The modernization the legacy pattern never produced: a cloud-native group administration platform with its database migrated to PostgreSQL and an IBM CDC bridge keeping DB2 in sync while the remaining mainframe dependencies are engineered out.
Five call types already handled on the AI contact center path, Spanish-language flows deployment-ready, and claims expansion in progress — member-facing value capture running today, not scheduled for a future phase.
Enrollment and premium reconciliation modernized as a platform lane: Recon Buddy phase 1 has been live since December 2025, and the ReconLink platform rebuild is in final UAT remediation with round-2 business signoff in motion.
05 — The economics
The business case moves the estate from $55.0M to a $26.5M target annual run rate — $1.28 to $0.60 per member per month. The discipline that keeps it credible: savings are counted when workloads are verifiably retired, not when code is migrated.
Source basis: WHPS AI Transformation strategic business case, January 2026. Full financial model and retirement ledger: financial view.
06 — The path forward
The program has proven it can build regulated, cloud-native platforms. What remains is engineering the mainframe out from underneath them — payment processing, batch, and the data bridge — and deciding who does which part of that work.
A proposed five-play doctrine: extract the business rules, flip data sovereignty domain by domain, decompose the batch estate, carve out payment processing, and gate every cutover on dual-run parity evidence.
A vendor modernization proposal validates the program's approach — pilot-first, modular decomposition, DB2 and VSAM to PostgreSQL. The assessment underway: self-perform the application build we have already proven, and engage vendor capability only where it closes a real gap.
Decision velocity now determines scale.
07 — The program record
Nine in active delivery, four decision-dependent, three queued, two live in production. Register dated July 1, 2026.
100% traceability coverage across 21 rows and 13 FIT cases; 2,337 published evidence files behind a review boundary.
Architecture views, security documentation, financial models, and every prior program surface, organized by question.
WHPS Transformation team — Led by Sam Sweilem.