Product strategy · Design · Evaluation
AI product design.
I connect customer needs, working prototypes, and evaluation results to shape useful AI experiences.
As product director at Ontraport, I work across product direction and experience design in collaboration with engineering and QA.
Start with the customer problem
The starting point is what a customer needs to understand or accomplish.
In a complex business platform, customers need help understanding features, finding the right place to take action, and making sense of their data. Those needs give the work a concrete purpose: help people get useful answers without interrupting their work.
I use that context to frame product decisions, explore interactions, and assess whether an answer actually helps someone move forward.
Build trust before expanding scope
Answer quality and clear expectations come first.
My approach is to establish confidence in the answers before expanding what an AI experience can do. People should understand its boundaries and know what to expect when they ask for help.
That means treating usefulness and reliability as product questions alongside interface design. A convincing response is not enough if it does not address the customer's need.
Make evaluation part of product development
I grew evaluation coverage from 40 to about 600 scenarios.
I helped build the evaluation program so model and prompt decisions could be informed by test results. Evaluation and in-app feedback give us evidence to discuss with engineering and QA as we refine the experience.
The scenario count reflects the breadth of evaluation coverage. It is not an accuracy score or a claim about customer adoption.
Balance quality, reliability, and cost
A useful AI experience has to work beyond a compelling demo.
I weigh answer quality, reliability, operating cost, and the flexibility to change models together. Evaluation results help inform those tradeoffs, alongside feedback from the people using the product.
My role is to keep those decisions connected to the customer experience and make the tradeoffs clear in conversations with engineering and QA.
Use prototypes to sharpen the spec
Working examples make product decisions concrete.
I work in design tools and code to explore interactions, test ideas, and clarify requirements. A prototype gives the team something specific to review and helps inform engineering implementation.
Across my product work, introducing AI prototyping into the spec process reduced the time from idea to spec by more than half.
Bring delivery experience to AI
My approach draws on more than 50 shipped projects at Ontraport.
That work spans payments, CRM data, booking, messaging, and integrations. It keeps my AI work grounded in the same responsibilities: understand the problem, make clear decisions, collaborate through implementation, and use feedback to decide what comes next.