CASE STUDY · ENTERPRISE AI · BANKING
Enterprise AI Knowledge Assistant for Banking Operations
Designing an AI-powered assistant that helped banking employees find answers 4× faster by replacing manual document search with conversational AI.
PRODUCT SNAPSHOT
Answers that explain the chart—not just retrieve the slide.
A contextual response combines the answer, data summary, source evidence, and next questions in one view.
ROLE
Lead UX Designer
CLIENT
Leading Private Bank · CDI
TIMELINE
12 Weeks · Enterprise Web App
BUSINESS IMPACT
4× Faster
Information retrieval
80%
Reduction in support tickets
20%
Increase in customer satisfaction
80%
Staff redeployed to higher-value work
01 · THE OPPORTUNITY
Find the right problem before designing the answer.
CDI wanted to explore how Generative AI could improve employee productivity across its operations. Employees across business, credit, and branch teams relied on thousands of documents every day—opening files, searching lengthy PDFs, interpreting charts, and contacting support when answers stayed hidden.
Instead of building another search tool, identify where employees were losing the most time, which problem AI should solve first, and whether the pilot could prove measurable business value.
Design Thinking Workshop
Mapping assumptions, knowledge flows, and opportunity areas.
RESEARCH ACTIVITIES
8
Business leaders interviewed
15
Banking employees interviewed
18
Cross-functional stakeholders
120+
Banking documents reviewed
02 · DISCOVERY
Three user groups emerged.
Rather than designing for everyone, research revealed distinct employee groups with different knowledge needs—and different moments where AI could reduce friction.
BUSINESS ANALYST
Extract insights from analyst presentations.
Pain points: lengthy presentations, charts, tables, and manual reports.
KNOWLEDGE BASE · Financial Analyst Presentations
CREDIT OFFICER
Find the latest credit policies.
Pain points: multiple circulars, scanned documents, and linked references.
KNOWLEDGE BASE · Credit Circulars
BRANCH MANAGER
Resolve core banking issues.
Pain points: raising tickets, waiting for Helpdesk, and delayed customer service.
KNOWLEDGE BASE · CBS Screenshots
03 · PRIORITISATION
Prioritising the right problem for a first pilot.
We evaluated each opportunity across business value, user impact, technical readiness, available data, and pilot feasibility.
PHASE 1 SELECTION
Business Analyst selected: highest business value, user impact, technical readiness, available data, and pilot feasibility.
04 · RESEARCH DEEP DIVE
Meet Rohit, the business analyst.
Before every leadership meeting, Rohit manually reviewed analyst presentations to extract financial performance, trends, and business insights. A single presentation could contain more than 180 slides of graphs, charts, tables, and commentary.
ARTIFACT · CONTEXTUAL INQUIRY & JOURNEY MAPPING
01 · Download presentation
02 · Read 180 slides
03 · Interpret graphs
04 · Compare tables
05 · Summarise findings
06 · Create PPT
07 · Present to leadership
Takes 2–3 hours · Multiple documents · Manual graph interpretation · Error-prone
05 · WHAT RESEARCH TOLD US
Searching, understanding, and trusting were inseparable.
FROM 15 INTERVIEWS
“I spend more time searching than analysing. Graphs take the longest to interpret—and every presentation has a different structure.”
73%
manually searched multiple documents
68%
struggled extracting information from charts
82%
wanted natural language search
2–3 hrs
average analysis time per presentation
06 · VALIDATED SOLUTION
A trustworthy answer, not another search result.
Before detailed UI, I tested interactive prototypes with realistic banking tasks. Iterations introduced document citations, structured answer cards, page references, chart summaries, and clear confidence cues.
Conversation screen
Natural language prompts, suggested questions, and follow-up context.
Visible evidence
Page numbers and document citations directly below every answer.
Complex content, explained
AI-generated chart summaries, table extraction, and concise key takeaways.
BUSINESS OUTCOMES AFTER THE PILOT
Faster access to banking knowledge translated into stronger customer outcomes.
4×
faster access to banking knowledge
80%
fewer branch support tickets · 2,000 → 400/month
80%
of associates freed for higher-value customer interactions
20%
improvement in customer satisfaction and NPS