AI Product Design Case Study
AI system for business automation
Overview
Turned “AI agents” into a product people can understand and use.
Problem
AI tools look powerful, but most people don’t get real output from them.
At enso, users were:
unsure what to do next;
clicking around without results;
dropping off before getting anything usable.
Constraints
Premium repositioning while keeping existing users. Engineering velocity that wouldn't tolerate per-feature design review. High-stakes flows (payments, agents, onboarding) where a bad call costs real revenue.
As well as:
Product was evolving fast;
AI output isn’t always predictable;
Desktop worked well, mobile didn’t;
Too many features competing for attention.
What was broken
We had a lot of capability, but no clear path.
Users had to figure things out themselves:
too many choices;
no clear flow;
no sense of progress.
Design decisions
We stopped thinking in screens and started thinking in outcomes.
simplified the main flow: generate > refine > use;
removed or hid low-impact options;
made the next step obvious at every point;
focused on getting users to a “first good result” fast.
Outcome
~10× ARR growth;
Trial-to-paid conversion nearly doubled;
Weekly active paid users grew ~10× (~100 → ~1,200+).