Fin'sAgent
Agentic Experience
Turning weeks of scattered research into minutes of trusted, sourced insight for analysts, AI -powered screening private market deals.
Context
My Role
Senior Product Designer
Responsibility
Research, Design, Prototyping, Testing
Duration
7 Months (Ongoing Iterative Evolution)
Teams
AI Engineers, Data Scientists, Product Stakeholders, and
Investment Domain Experts.
Impact
“Impact” — Transforming Financial
Operations
Agent Fleet Delivers Speed, Accuracy, and Sustainable Impact
90%
Reduction in time spent on first drafts
75%
Accuracy in initial outputs
50%
Reduction in operational costs
2M+
Market sources analysed instantly
100+
Workflows supported
10x
Faster investment research and reporting
#Pain Point
Critical data scattered across
hundreds of unverified sources
No fast way to establish source
credibility before acting on it
Insight generation required
deep synthesis across
unstructured documents
Time-sensitive screening
cycles left little margin for this
manual overhead
Before
Manual Work-flow
01
Data Hunting through multiple sites.
02
detailed excel sheets to input
03
process it manually
04
perform multiple validation checks
05
Upload to respective database.
Solution
Solution We came up with....
We reimagined the pre-investment workflow as a collaboration between analysts and intelligent agents.
Fin'sAgent was designed as a prompt-based ecosystem of AI agents that can gather, consolidate, and
structure information at speed while maintaining transparency and human control. Instead of asking
analysts to adapt to AI, we structured the experience around their natural workflow.
Service Design
“Agent Fleet” – Human + Tech
Redefining collaboration, where human expertise meets AI precision
Financial World
Acuity Analyst
Agent Fleet
Experience
The Perfect Handshake between
Humans and AI
Process
Process that insurance the accuracy of the output
This is the core of Design where this works as learning steps for designer also to Understand about the
User, Business Gole, Journey, Problems
01
Understand the existing process in depth
02
Conduct stakeholder interviews to better understand pain points and needs
03
Performed Competitive research and benchmarking
04
Define opportunity areas from a product perspective
05
Craft major user journeys and information architecture to kickstart the design
Findings
What we Uncovered ?
Users interviewed
25
Products analysed
7 internal products
Competitors researched
Alphasense, Rogo, Hepia etc.
Major Finding
The company has over 120 workflows (most of them
carried out manually), which can be automated using a
similar AI journey
Strategy
The work-flow behind the Interface
The Strategy that helped us Shape the Output
01
Modular Experience for Diverse Needs
02
Human in the Loop AI Framework
03
Progressive Disclosure of Complexity
04
Templates, Defaults and Configurable System
Final Shipped Ideation
The Core Idea that made this Product stand
out
Configurable by Design — Templates
That Fit Every Need
Admins build reusable prompt and publishing templates, so end users can
generate and present AI outputs instantly—no setup, no fuss. It's fast,
flexible, and built to scale.
Configurable by Design — Templates That Fit Every Need
Run Time Agents Solving Real
Problems
Once set up, agents run the workflows automatically, pulling data,
analyzing it, and giving results in real time. The interface clearly shows
what's happening, so users can track progress and outputs easily.
Run Time Agents Solving Real Problems
Built-In Trust Through Source
Visibility
Every insight surfaced by Fin'sAgent. is accompanied by clear source
attribution and an accessible audit trail. Analysts can instantly validate
where information originated From and they can audit Source of
Information.
Built-In Trust Through Source Visibility
Ideation That Didn't Land
Not everything we create ships, but every unshipped idea taught us something. The final
version wouldn't exist without them.
Ideation - 01
Freeform open chat (ChatGPT-
style input)
The very first instinct for an agentic AI
product in 2024-25 was an open text box
type anything, get anything. It's the pattern
users already know.
Freeform open chat (ChatGPT-style input)
Ideation - 02
The Floating AI Chat Assistant
An early direction placed the AI assistant as
a floating, dismissible chat widget in the
bottom-right corner accessible on demand,
sitting on top of the workspace, similar to a
support-chat pattern.
The Floating AI Chat Assistant
Ideation - 03
Manual re-prioritisation of the
agent queue
Our obvious first idea is to let the analyst
reorder or reprioritise which module the
fleet tackles first e.g., drag Financial
Commentary above “News” because that's
the one they actually need in the next five
minutes, rather than waiting for the system's
default sequence.
Manual re-prioritisation of the agent queue
Outcome
Life Before Fin'sAgent.
Manual data extraction from multiple fragmented sources
Switching between 6–8 platforms to complete one
investment analysis
Delayed insights due to time spent on cleaning and
structuring data
Limited visibility into multi-scenario projections
Redundant workflows increasing cost per analysis
Life After Fin'sAgent.
AI agents automatically collect, clean, and structure data
Intelligent summaries and anomaly detection
Real-time cross-validation across datasets
Predictive risk signals and scenario simulations
Reduced operational cost
Reflection
Looking Back : What I'd Do Differently
Research beyond Acuity.
Fin'sAgent was shaped around the workflows of in-house Acuity analysts sharp and focused for that
team, but never tested against how analysts elsewhere actually work.
Test trust from the very first
version.
Design for more than the
expert user.
Fin'sAgent assumes a baseline of financial fluency. I'd want to test it with people newer to investment
evaluation too to see how much the model could actually teach, not just accelerate.
Key Learning
Empower Humans, Don't Replace Them
The strongest impact came when AI handled repetition, allowing analysts and fund managers to focus on strategic judgment and decision-making.
Designing for the ecosystem is the designer's
superpower
I realised this kind of product needs design thinking beyond the screen, considering every user, workflow, and operational touchpoint.
Fin'sAgent
A Case- Study by Divya Bhushan Singh