Enterprise engagement · Fintech · Enterprise analytics
FinSight AI — From fragmented finance to real-time intelligence
An AI-powered financial intelligence platform helping CFOs monitor company health, predict risk and decide faster
- Faster monthly close
- 68%
- Data sources connected
- 4.2×
- Usability score
- SUS 87
- NPS after 90 days
- +22
Overview
Reports took days to compile, controllers surfaced raw data, and executives decided on stale numbers. FinSight AI consolidates fragmented sources and turns a periodic reporting exercise into an always-on view of company health.
A three-month discovery combined desk research, competitive teardowns of Ramp, Mosaic, Cube and Anaplan, and shadowing two finance teams through their monthly close.
Discovery
Finance leaders do not lack data — they lack trustworthy, timely insight
Desk and market research mapped the tooling landscape. Contextual enquiry shadowed a live monthly close. A data audit of sample ERP and banking exports exposed the structural mismatches that make reconciliation manual.
Across 14 conversations one theme recurred: “I don't present a number I can't trust.”
- Data lives in 5+ tools with no single source of truth
- Spreadsheet sprawl, version conflicts and manual drift
- Reporting is manual, periodic and error-prone
- Executives need answers in the moment, not next week
Roles
Four roles, four very different jobs-to-be-done
Alex, CFO — growth and risk, needs where we will be next month. Filipa, Finance Manager — a close that takes days. Marcus, Analyst — ad-hoc questions answered with confidence. Sofia, Controller — audit trails and anomaly detection.
Information architecture
Structured around decisions, not accounting modules
Rather than mirroring the general ledger, the IA follows the questions finance leaders actually ask: how healthy are we, what is about to break, and where are we headed.
- Overview — health score, cash, revenue, burn, KPIs
- Cash flow & burn — inflows, outflows, runway
- P&L insights — anomalies, variance, plan vs actuals
- Headcount, revenue, payroll and expenses
- Data settings — sources, permissions, audit
AI experience
Ask anything. Get answers with receipts.
A conversational assistant grounded in the company's financial records surfaces anomalies, risks and opportunities without being asked — and every answer links back to the underlying data.
A risk matrix maps liquidity, operational and market risk with confidence levels so teams can plan mitigations instead of firefighting. Driver-based forecasting updates 12-month projections in real time with assumptions documented on screen.
Design decisions
One number executives learn to trust
The financial health score became the emotional anchor of the product: a composite of liquidity, profitability, efficiency, solvency and growth, benchmarked against peers and trended over 12 months with clear recommendations.
The system emphasises restrained colour, generous type hierarchy and 1px chart craft — high-density visualisation that stays legible. White space was never the goal; clarity was.
Testing & outcome
Three rounds, three hypotheses
Rounds targeted navigation clarity, trust in AI, and speed to insight: 94% task success, 47s time to first insight, SUS 87, and 9.1/10 self-reported trust in AI outputs.
After a 90-day pilot the monthly close shifted from a two-week manual process to continuous, automated cycles — 68% faster close, 4.2× more sources connected, +22 NPS.
Reflection
Trust is the feature
It was not enough for the AI to be accurate; it had to be easy to trust. Every output needed to be explainable, and the design system had to ship alongside the screens for the team to move at all.
Next case study
ShopSense AI — Shop smarter, personalised by AI
Commerce · OmnichannelContact
Let’s talk about the problem, not just the solution.
I’m open to senior product design, UX leadership, design systems, and AI-native product opportunities.
Get in touch