All work

Case study · Xceptor

AI-powered Recon builder

The north-star concept I designed for AI-powered reconciliation tested and scored SUS 80+. We shipped a deliberately smaller first test of that vision, and what it taught us is shaping what Recon builder does next.

Company

Xceptor

Role

UX Designer

Timeline

2022 – present

Team

End users, Internal SMEs, PM, PO, Engineers, Head of UX

Overview

Reconciliation teams spend most of their time on setup: mapping fields, creating rules including perfect match and tolerance match, before a single match gets looked at. My brief was to use AI to collapse that setup time without trading away analyst trust in what the model suggests.

What shipped

Outcome

I ran a comparative study against the north-star benchmark. Based on a timed run with an internal SME, the north-star approach could save config users more than 50% of the time spent on current manual reconciliation setup. Testing also surfaced a signal that existing reconciliation configurations benefit from AI recommendations as well, improving STP and reducing manual intervention. Testing surfaced visibility and trust as the key factors for the next phase, building an evidence-backed case for enhancing the product with fuller control through a few different approaches, now shaping the next stage of the roadmap.

CEO Recognition, 2025

AI-Enabled Reconciliation Workflow Design, Xceptor

Xceptional Award, 2026

GTM Growth Award, for AI-powered rule optimisation, Xceptor

Starting smaller than the vision was the plan, not a compromise. Ship the test, read what it tells you honestly, and let that evidence make the case for the next investment.