Case study
Speaking Coach
A speaking-practice app that runs entirely on my own computer. I record myself, it transcribes what I said and gives structured feedback against real communication frameworks, with no subscription and nothing leaving my Mac.
How it works
Record a practice session on my own computer
The recording is transcribed automatically, entirely on-device
A free tier of an AI model analyzes delivery and content
Feedback shows up in a simple web interface, with no cloud service and no subscription cost
Curriculum
A lesson library grounded in the communicators I actually study.
A lesson, opened
Opening a lesson gives you its source, a few action steps, and one drill to actually do.
What was actually broken
Practicing alone had no structure.
I already knew from studying other communicators that speaking out loud daily, on a random topic, cutting filler words, building a stronger connection between thinking and speaking, is one of the most effective ways to actually improve. There was no way to track progress over time, no single place to review past sessions, and nothing tailored to how I actually learn.
What I tried, and got wrong
What was actually missing was structure, not better recording tools.
I looked at paid tools and passed. It wasn't really about the cost, I knew I just needed to film myself.
What it does now
A speaking-practice tool that costs nothing to run.
I record, get structured feedback on delivery and content, and iterate, without sending anything to a paid service or a cloud account. It runs on the free tier of Google's Gemini model, which comes with usage limits, but far more than I need for daily practice.
What changed
That includes a daily practice habit with AI feedback, and tailored practice for specific situations like small talk and difficult conversations.