Distill
Why this needs to exist now
Written communication used to be the bottleneck for most knowledge workers. It is not anymore. Models can draft the memo, summarise the meeting, polish the email, and produce a passable cover letter in seconds. The lever that used to set people apart is being levelled out at machine speed.
What does not get automated is the part where you stand up in a room, real or virtual, and explain something so people actually understand and remember it. The pitch, the team meeting, the customer demo, the conference talk, the interview, the difficult conversation. Speaking is now the part of communication that humans uniquely own, and almost no one is deliberately practising it.
Distill is built for that gap. It is a speaking coach you can point at the talks you actually have to give, that returns the kind of feedback you would only otherwise get from a senior coach watching you closely.
Drill on the questions you'll actually be asked
Distill is organised around scenarios, not exercises. Pick the round you are preparing for (behavioural, technical screen, exec readout, customer pitch) and you get a deck of questions sourced from that context, each with a real rubric you can see while you answer. The rubric is the part that makes the practice honest: it tells you what a strong answer needs to cover before you record, and grades your take against the same criteria afterwards.
Feedback you can act on, not vibes
After each take, Distill scores the answer on the things that actually decide whether you came across well: coverage of the rubric, clarity, confidence, pacing, and filler density. Every metric is paired with specific notes about what landed, what to tighten, and what to add next time, so you know exactly what to change on the next take instead of just "try again, but better".
Why I'm building it
The bet is that as AI keeps absorbing the written and analytical layers of the work, the people who do well will be the ones who can stand up and be clear, persuasive, and engaging in person. That is a learnable skill, but almost no one trains it deliberately, because the feedback loop is so weak. You give a talk, someone says "nice job", and you have no idea what to actually change next time.
Distill closes that loop. The goal is to give every person who has to communicate for a living the same kind of structured, honest, repeatable practice that elite speakers have always had access to, except now it scales and lives on your laptop.