TrustPro
Tailored AI document extraction for financial institutions
Product Design
AI Integration
Design Engineering
↑
HERO ANIMATION — Split screen. Document on left, empty sidebar on right. Scanning motion moves down the document as fields populate in the sidebar. A field is clicked and the document jumps to the highlighted source passage. 8-12 seconds, looping.
Challenges
Financial institutions deal in complexity. Wealth managers, banking executives, and compliance teams regularly work with dense legal and financial documents, trust deeds, wills, constitutions, minutes of meetings, partnership agreements, each one potentially running to hundreds of pages. Extracting the information they need from these documents is a slow, manual process that requires not just careful reading but genuine legal and financial comprehension.
The problem is more challenging when portfolios have nested structures. For example, a trust can name another trust as a beneficiary, which can itself have further complex ownership. Tracing the ultimate beneficial owner across a web of related entities means cross-referencing multiple documents simultaneously, a task that could take an experienced professional hours and still leave room for error.
Existing tools were too broad to be useful in this context. Nothing had been built specifically for the workflows, terminology, and risk tolerance of financial institutions. For organisations that are among the most cautious adopters of new technology, that lack of fit was a genuine barrier.
KEY CHALLENGES
01
Extracting key data from dense financial and legal documents was slow and entirely manual
02
Tracing ultimate beneficial ownership across nested trust structures was complex and error-prone
03
No existing tool was purpose-built for the specific workflows of wealth managers and financial institutions
04
Risk-averse clients had strict requirements around data sovereignty and document privacy
↑
Each extracted data point is embedded with metadata at the point of capture, giving users a complete and traceable audit trail from source document to verified output.

↑
Bolt is TrustPro's ground-up design system, built in Storybook with accessibility as a core principle rather than an afterthought. It gave the engineering team a reliable, consistent foundation to build against throughout the entire product.
Solution
I joined TrustPro at the earliest stage of ideation, helping shape the product concept before any designs existed. From there I built everything from the ground up: brand, visual language, design system, and a significant portion of the frontend. I also built and maintained a Storybook design system, giving the team a reliable component library that stayed in sync with the product as it grew. The backend team handled the technical infrastructure and assisted with frontend development, but all the styling, components, and UI decisions were mine. I was also in the room during informal sessions with wealth managers and banking executives, and those conversations directly shaped the product at every stage.
TrustPro Verify was built around one clear principle: AI surfaces the information, the document proves it.
The Verify workflow
01
Extract
02
Verify
03
Chat
04
Export
A user uploads a suite of documents, a trust deed, a company constitution, minutes of meetings, and TrustPro processes them together, understanding not just what each document says but how they relate to each other. The extracted data populates a structured sidebar, giving the user an immediate picture of something that would previously have taken significant time to piece together manually. For complex portfolios with nested trust structures, the ability to process multiple documents as a connected whole was the product's most meaningful capability.

Credits
Role
Product Design Lead
Client
TrustPro
Year
Team
Andrew Twaits
Gus Gollings
Steve Hodgkiss
Anthony Novatsis
Tim Andrews
Arman Mirkazemi
Myka Hecht-Wendt
