Experiments
Blindspot
Built a working AI product in days that turns unstructured customer feedback into prioritized product gaps and draft documentation.
Why this experiment matters
Most product teams are not short on feedback. They are short on clarity.
Feedback lives everywhere: support tickets, reviews, docs, Slack threads, customer calls. The problem is not access. It is synthesis.
I built Blindspot to ingest these fragmented signals, identify what actually matters, and generate actionable outputs including prioritized gaps and draft documentation.
This changes the workflow from collect and read to analyze and act.
Experiment
Context
Modern product teams have signal everywhere, but clarity nowhere.
Modern product teams operate in environments where customer signals are abundant but fragmented. Documentation exists across help centers, APIs, and internal systems. Feedback exists across review platforms, forums, support tickets, and conversations.
The core problem is not lack of data. It is lack of synthesis.
Teams struggle to answer what the most critical gaps in their documentation are, where customers consistently get stuck, what should be fixed first, and how to move from signal to action without manual analysis.
The result is slow iteration, reactive improvements, and missed opportunities to improve onboarding, retention, and product experience.
What I built
I designed and built Blindspot as an end-to-end system from signal to output.
Built an ingestion layer that accepts documentation URLs and feedback sources, then crawls and normalizes both into a unified input corpus.
Built signal processing and AI analysis to identify recurring themes, cluster related issues, and detect friction points across the customer journey.
Built a gap prioritization system that ranks issues based on frequency, severity, customer impact, and documentation coverage gaps so teams can move from everything is important to fix this first.
Built a draft generation layer that transforms insight into draft documentation, onboarding guides, and suggested improvements, reducing the gap between insight and execution.
Outcome(s)
From fragmented inputs to prioritized action.
Functional vertical slice built in days.
End-to-end system from ingestion → analysis → output.
Demonstrated ability to reduce manual analysis, prioritize work, and accelerate documentation improvement.
Why it matters
This is not just an AI experiment.
It demonstrates how I think about systems: inputs → signals → prioritization → output.
It connects data to action and reduces operational friction.
Most teams stop at insight. I build systems that produce outcomes.
Related Experiments