Our Story
The journey that brought us to create UserTrace - A platform designed to help teams simulate and evaluate AI agent before launch.
It Started with a Shared Frustration
While building and deployed AI products in compliance-heavy domains like healthcare and finance, and seen firsthand how broken evaluations, safety risks, and regulatory gaps kill products.
There was no reliable way to test how AI agents behave in realistic, multi-turn user journeys before launch.
Testing AI-human interactions felt like flying blind.
Our Insight
At Dheeraj's previous startup (Zealth AI), he spent a year securing a pilot with a top Indian hospital for our remote cancer monitoring solution. In the first week, one patient received a response that didn't align with the doctor's protocol. The pilot was shut down immediately, and lost over year of effort.
After that, spent weeks manually crafting edge cases before every single prompt update, knowing that one failure could kill trust.
At CleanHub, Deepali hit the same wall we've seen everywhere: manual evaluation that couldn't keep up with changing models and growing usage.
Realizing We Were Not Alone
We've spoken to 80+ AI PMs and engineers across SaaS, healthcare, and BFSI. The pain is the same:
"How do I test what will break before I go live?"
What We Heard
💬 "We are creating sandbox environments just to know what changes to expect — if your MCP shows impact before production, that's golden."
— CTO, Oncology AI Agent, US Healthcare
💬 "Could this be plug-and-play for AI regulations, bias testing… that would be a relief."
— Product Lead, Mental Health AI
💬 "Multi-turn scenarios are manually created… the dev wants to try new ideas, but the cost of evaluation slows them down. If you can reduce that to hours, that's a win."
— VP Product, BFSI Enterprise
💬 "This is exactly the kind of simulation layer we're missing. I love the part where you eliminate the need to think through every edge case. That's the biggest inertia we face."
— Senior PM, BFSI Startup
The Birth of UserTrace
With UserTrace, we're combining user insight and technical depth to build a simulation-first evaluation platform for safer, faster AI deployment.
Our Mission Today
At Usertrace, we belive that every AI deployment in a high-stakes domain will require simulation-first validation. We are building that infrastructure.
Our mission:
- Simulate real-world journeys across personas, workflows, and edge cases.
- Run automated evaluations for safety, compliance, and performance.
- Surface failures early so teams can launch with confidence.
Meet the Team
We've spent the last decade building and deploying high-stakes AI systems
Dheeraj is a machine learning expert who has built AI products across healthcare, fintech, and e-commerce.
He was the first engineer at SigTuple, where he led the development of an AI-driven blood diagnostics platform, secured three granted patents, published research papers, and helped deploy the solution across 20 laboratories in Asia.
Later, he co-founded Zealth AI, a remote patient monitoring platform for COVID and cancer care, implemented with top-tier hospitals like AIIMS, Apollo, and Medanta. The platform supported over 20,000 patients and improved clinical outcomes through real-time symptom monitoring and escalation.
Deepali is an experienced product leader specializing in building 0-to-1 products across B2B and B2C supply chains.
At Flipkart, India's largest e-commerce company, she launched and scaled Shopsy, a hyper-value commerce app that reached over 80 million downloads and generated $7 billion in GMV within 18 months, by solving affordability gaps in tier 2 and tier 3 cities.
Most recently, as a Senior Product Manager at CleanHub, she built traceability software and an AI auditor for waste managers and recyclers, helping brands achieve global sustainability targets through real-time supply chain monitoring.