
1
1
United States
1 year of experience
Swim coach for 10+ years. I build AI tools for swim schools and families: CoachCover, BillRadar.

I coached swimming for more than ten years. The worst mornings began with a coach writing at 08:40 that they couldn't make the 09:30 class, and the manager phoning around for someone with the right certificates who was cleared for that program. CoachCover is an agent that does that job. A coach sends a voice note in Russian, Spanish or English. Whisper on Vultr turns it into text, Muse Glimmer translates it, and the agent finds the lessons it affects. The code decides who may take a lesson: role, every required certificate and its dates, program and level, weekly hours, rest between days, overlapping bookings and the children-per-instructor ratio. The models only transcribe, translate, read cards and write messages. Offers go to people the rules allow, three at a time, and the rules run again when someone says yes. For a private lesson the parent is asked first. If they say no, the lesson moves to the child's own coach, or the family gets a credit. When nobody can come, the manager gets plan B (reserve coach, overtime, move or cancel). Approval is tied to a hash of the facts, so a plan whose facts changed is refused. A school can borrow a coach from a sister pool over A2A once both managers agree. A manager's AI assistant can ask who is allowed and request cover over MCP, but approving plan B or answering for a family is refused by policy. A nightly check opens cases before the morning, and the inspector's folder is a PDF one click away. It all runs on Vultr: one VM for the app, the agents, the MCP server and Postgres, and four models on Vultr Serverless Inference. The scenes call Whisper, Muse Glimmer and Nemotron Safety. Nemotron Omni reads certificate cards together with Muse Glimmer, live through "Check a sample certificate" on the Staff page. We replayed 50 synthetic mornings. With a group chat and phone calls, 45 of 98 replacements went to someone who wasn't cleared. With CoachCover, none did. In the replay, how families answer is assumed, not measured.
8 Nov 2026