Location
NYC
Timeline
Jan 2024—May 2024
Role
Lead Product Designer
Team
4 Designers, 1 PM, 4 Devs
THE STORY
AI is powerful, but people don’t trust what they can’t understand.
At TextQL, a $4.1M seed-stage startup, I redesigned Ana, an AI data analyst that translates natural language into SQL queries. The goal was to help users trust AI-generated insights without losing control or clarity.
THE Solution
Making data work feel natural, not technical.
Ana allows analysts to ask questions like “What were our top campaigns last quarter?” and instantly get visualized results. I redesigned Ana’s dashboard to make that interaction feel simple, structured, and transparent, so AI felt like a teammate, not a mystery.
The solution we came up with that got shipped!
Translating queries
Ana translates the query into SQL, runs it, and returns clear visuals, summaries, and insights directly in the dashboard.
Personal Data
Ana acts like a personal data scientist, cleaning data, finding correlations, and surfacing outliers, all while keeping users in full control.
Product showcase
Ana
Ana is an AI data analyst that translates natural language questions.
THE OUTCOMES
Product shipped!
After rollout, the redesign led to measurable improvements such as 40% faster data exploration which streamlined navigation and clearer workflows
My Individual Contributions
Product Design, PM Collaboration, User Research
80+ final screens across Ana’s redesign
Led usability testing and synthesis
Defined design system
That was the TLDR; continue below for details!
Process Overview
Timeline & Roles
Problem definition
Why Analysts Distrusted Ana
After 25+ interviews with data professionals from Google, PNC, JPMorgan, and Shell, we uncovered three recurring pain points:
Poor conversation context
Inefficient navigation
Overwhelming data presentation
Secondary Research
Understanding AI in Enterprise Data Workflows
We studied how large organizations used AI for analytics. Findings revealed a consistent truth: speed without transparency leads to distrust. This insight shaped our north star, make Ana understandable, not just intelligent.
Design considerations
Poor Conversation Context
The original Ana made it difficult to track which questions were answered or still open. I introduced threaded conversations and contextual suggestions like: “Show me a visualization,” “Add to dashboard,” or “Let’s keep going.” This guided users through analysis naturally, improving flow and comprehension.
Inefficient Navigation
The old design had two overlapping sidebars that confused users.
I restructured navigation into a single unified system organized by threads.
Now, users could seamlessly switch between analyses, view history, and search results, without losing their place.
Overwhelming Data Presentation
Analysts were overwhelmed by unformatted data. I redesigned the layout with clear visual hierarchy, expandable result sections, and hover-based details for SQL logic. This created a balanced experience, comprehensive yet approachable.
Responses feel disconnected
Hard to track previous insights
Not intuitive
Navigation on both sides
Too much info at once
Difficult to prioritize insights
Ideation
Round 1
The original Ana made it difficult to track which questions were answered or still open. I introduced threaded conversations and contextual suggestions like: “Show me a visualization,” “Add to dashboard,” or “Let’s keep going.” This guided users through analysis naturally, improving flow and comprehension.
Design A
Design B
Design C
100% of users notice the design,
but 80% admitted they did not read
the information
Only 30% found it engaging enough
to continue further
65% of users noticed and
understood more effectively
Some users mistakenly thought the
feature signaled the end
PIVOt
Changing our choices
Mid way through our initial ideation. we were faced with a mid contract pivot where I had to work with the engineering and pm teams as a leader because of
Real-time dynamic previews caused performance issues
Negotiated a pivot → “preview-on-demand”
Preserved clarity, improved performance
Ideation
Round 2
Post pivot, I was tasked by the client to come up with drastically different ideation that now incoropartes the feedback we got from the developers and pms. I then came up with the following designs and user tested them to see which would be the most optimal for this platform.
Design A
Design B
Design C(Winner!)
70% of users preferred this design
for its clean concise layout
20% wanted more information
visible upfront
30% preferred this design for its visibility
60% of users felt this was cluttered and occupied excessive space
Final Designs
Before & After
The once cluttered double navbar, now has a simplifed interface, where users can access the main point of the platform easily, which is Ana.
Before
After
Navigation
By gathering data on what technical users actually ask, the conversation flows better due to now prompted follow-ups.
Before
After
Conversation Context
Because of cluttered UI, among other factors, the once unreadable data presentation has now been structured to format the files given.
Before
After
Presenting Data
end Credits
The warmest thank yous
This project was one of my first ever shipped projects. Leading a team of 4 designers while working with AI was such an incredible experience. It truly taught me so much about working with engineers and pms along with other designers as well. I am so grateful to be part of such an innovative product and am happy with all that I have learned from it.