Building a continuous discovery engine for an AI investing app
Uncovering unmet user needs through weekly interviews to define ICP and inform the product roadmap
An AI investing app without a clear customer
Rafa.ai is an AI-powered investing copilot built to help users make smarter financial decisions. When I joined as a Product Discovery Consultant, the team had a compelling product but an unclear picture of who their ideal customer was and what would keep them coming back.
My role was to build a structured, ongoing discovery program from scratch by running weekly user interviews, mapping insights to product opportunities, and feeding those directly into the roadmap alongside the product and engineering team.
Three things to answer through research
- Define the ideal customer profile (ICP) who they are, what they need, and their biggest pain points
- Use those needs to streamline the first-time user experience toward a clear desired outcome
- Uncover opportunities for ongoing engagement loops, brainstorm solutions, and test assumptions in-product
A three-channel pipeline for discovery calls
Before any interviews could happen, I needed a reliable way to recruit participants consistently. I designed three recruitment channels to run in parallel:
Weekly interviews → weekly insights → product direction
Each week I conducted 1:1 interviews with current and prospective users, focusing on uncovering unspoken needs, daily habits around investing, and friction points with the existing app. Insights were synthesized into an opportunity map that I presented to the product, engineering, and design team weekly.
The opportunity map connected each interview insight to a user opportunity, a potential product solution, and an in-product assumption test which keept the team focused on what was worth building and what needed validation first.
From insights to in-product tests
Rather than waiting for a full feature build, I worked with the team to design small, measurable in-product tests for each assumption. Weekly analytics reviews tracked test performance against engagement and retention health metrics which created a tight loop between what users said in interviews and what the data showed in the product.
Continuous learning that drove product-market fit
The discovery program ran throughout 2023–2024, producing a steady stream of user insights that directly informed features built, pivots taken, and the first-time user experience. Rafa.ai has since reached product-market fit.
Discovery is a system, not a one-time event
The most valuable thing about this engagement wasn't any single insight it was the cadence. Weekly interviews created a compounding understanding of the user that no one-time research sprint could replicate. The opportunity map gave the team a shared language for connecting user needs to product decisions, and the in-product tests kept assumptions honest.