SKIP TO CONTENT
← ALL WORK
01 — SPARQ · FINTECH · 2023

Board-meeting prep for bank executives, cut from hours to minutes.

SPARQ is an iPad workspace for banking managers. It summarises the documents that arrive before a board meeting, keeps market sentiment on the same screen, and writes up the minutes everyone shares afterwards.

Role
Lead UX/UI — research, IA, UI, usability testing
Client
Userdata
Timeline
2023 · 12 weeks
Platform
iPad
SPARQ — executive briefing
−72%
prep time per board meeting, reported by pilot managers
1 tap
from any AI summary to its source passage
6
core screens shipped, dark-first design system
2
usability rounds before engineering handoff

Problem

Managers used to prepare for a fortnightly board meeting by copying figures out of email attachments by hand. Shareholder and sentiment data sat in other tools, so they went looking there too, then wrote the minutes afterwards from memory. Prep ran to hours, the numbers were easy to get wrong, and nothing survived the meeting that anyone could audit.

In interviews the same three complaints came back. Finding the right document took too long. The tools they had were years behind the job. And on the morning it mattered, there was no quick way to read the market.

Research

I shadowed six managers for two weeks.

I sat in on real prep sessions and mapped every switch between tools. The average manager opened 7 sources before a meeting and re-typed the same figures three or four times. Everyone I interviewed said a version of the same sentence: “I don’t need more data. I need it in the room.”

7
tools touched per prep session
3–4×
same figure re-typed by hand
0
audit trail for decisions made

Process

  1. DISCOVER
    Stakeholder interviews, prep-session shadowing, competitive teardown of Bloomberg-style tools
  2. DEFINE
    Meeting-centric information architecture; success metric = minutes to "ready"
  3. DESIGN
    Wireframes → dark-first UI system → coded prototype with real documents
  4. VALIDATE
    Two usability rounds with managers; summary-trust issues fixed before handoff
  5. SHIP
    Token-based design system delivered; engineers still use it

Decisions

01

Build everything around the meeting.

Every document, metric and note attaches to one board meeting, so prep is a single screen instead of a search.

TRADE-OFF

Browsing the document library on its own got slower. I took that trade, because prep is the job the tool exists for.

02

Every summary shows its evidence.

A summary lists what it pulled out, and one tap opens the passage it came from.

TRADE-OFF

The screens carry more than a clean summary would. Executives will not act on a figure they cannot check, so density was the cheaper problem.

03

Design dark-first for long reading on iPad.

Financial tables and charts were re-tuned for a dark palette that holds contrast through a two-hour meeting.

TRADE-OFF

The palette took far longer to settle than a light one would. It also became the thing people recognise the product by.

PrototypeTap the glowing hotspots to move through the app, like the real prototype.

SPARQ — Executive briefing
SPARQ — My applications
SPARQ — Application detail
SPARQ — Market sentiment
SPARQ — Key financials
SPARQ — Meeting summary
Executive briefing
The meeting-centric home: every document, metric and note attached to the next board meeting.
01 / 06

Before → After

BEFORE

Email attachments, PDF exports and a legacy intranet. Figures re-typed by hand, minutes written afterwards from memory.

AFTER

One iPad. Documents are summarised as they arrive, sentiment sits beside them, and the minutes go out before anyone leaves the room.

Outcome

  • SPARQ pulls the key figures out of every financial document, so nobody copies them by hand.
  • Shareholder data and market reputation read at a glance, on the same screen as the documents.
  • Meetings run off live screens and end with minutes ready to share.

Reflection

The AI was the easy part. Getting an executive to act on a number a model produced took much longer, and no summary was worth anything until it sat one tap from its source. I should have measured time-to-ready from the first week, and tested the summaries against the notes managers still write by hand.