Financial Services

How Neoware Built an Audit-Ready Rules Engine

AI/MLGen AI

Client Overview

A prominent UK pension administrator carrying decades of institutional knowledge. This firm administers complex pension schemes for a large customer base where accuracy, trust, and data integrity are non-negotiable. Modernizing with GenAI is no longer an efficiency play – it is a competitive necessity to win new customers.

The Problem

Every customer has a set of trust deeds and scheme documents — dense legal and actuarial PDFs running 500 pages each. Across the client’s full portfolio, roughly 2,000 such document sets exist, layered with decades of shifting pension legislation notifications.

Pension quote requests for retirement, divorce, or bereavement require precise & consistent calculations
Teams manually cross-referenced trust deeds, scheme documents, benefits summaries and historical legislation to determine the right rules
Each request took 35–40 days, driven by complex manual rules extraction and document review
A 2,000-person operations team handles this manual workload
The cost of error is significantly high – in terms of effort to fix and suboptimal customer experience
Four vendors had already failed to solve this complex, highly validated business challenge

The Result: 35–40 day turnaround times, high manual effort, and significant accuracy risk, slowing critical pension decisions at scale.

The Solution

Neoware built an LLM-based solution to process scheme documents, historical legislation, and trust deeds to automatically generate the Control Workbook (CCW), the foundation for determining the rules applicable to each member’s pension calculation. The solution was designed for accuracy, traceability, and human approval, with validation across 20 pension customers before portfolio-wide rollout.

Financial Services

Phase-1 – Automated Control Workbook Generation
The system processes scheme documents, legislation, trust deeds and benefits summaries for each customer and builds a graph+ vector database to produce a completed 110-question Control Workbook automatically, leveraging a combination of semantic search and graph RAG.
Full Traceability by Design
Every answer in the Control Workbook is linked back to the exact source document, page, and sentence it was derived from giving operations teams an audit trail that they can rely on when a member later questions a rule applied to their case.
Proven Before Scaling
The solution was validated across 20 of the client’s pension customers ahead of full portfolio rollout, with a clear path defined for Phase 2, automating the downstream quote generation itself, with a human-in-the-loop review and approval step preserved to satisfy regulatory requirements.

The Business Outcomes

Solved What Four Vendors Couldn't

Delivered a working, production-track solution to a business problem the client's leadership had explicitly flagged as unsolved after multiple prior attempts.
Control Workbook Generation: Days to Minutes

The CCW step previously around 26 to 27 days of the total cycle is now produced through a fully automated process in a matter of minutes.
35–40 Days Down to 5 Days

Phase 1 alone has yielded a productivity saving of greater than 75%, even before the remaining manual quote-generation step is automated.
Path to a 2–3 Day Turnaround

Once Phase 2 is complete, the full process is projected to run in 2 to 3 days with the remaining time attributable almost entirely to queued human review, not processing.
Proven at 20 Customers, Built to Scale to 450

Validated and structured to scale across the client's full base of pension schemes.
A 2,000-Person Team Redeployed, Not Reduced

Freeing operations analysts from manual document review to focus on higher-order, higher-value work for the client.

The Journey From Data to Intelligence: How Neoware Helps

Neoware helps pension and benefits administrators tackle the problems other vendors leave behind, turning decades of legacy scheme documentation and shifting legislation into an automated, auditable, rules-based engine. Where four prior attempts stalled, Neoware built a solution accurate enough to trust, traceable enough to defend, and structured to scale across an entire portfolio, not just a handful of pilot customers.

About Neoware

Neoware is a specialist Data to Intelligence solutions provider combining consulting, data engineering, GenAI, AI/ML, BI and Data Privacy, Compliance, & Governance expertise with proprietary accelerators to deliver measurable business outcomes. Recognized as a Top 30 AWS Data-Driven Everything Partner and a Seasoned Vendor in AIM Research’s PeMa Quadrant for both Top Data Engineering Service Providers 2026 and Top Generative AI Service Providers 2026, with leadership recognized among the Next 15 AI CEOs in America 2026, we build secure, scalable data platforms with GCP, Snowflake, Databricks, and Mage Data partnerships.