
It listens, without a bot
Records both sides of the call from your own computer. Nobody sees a guest join.
It listens without a bot, finds what was agreed, does the work in your tools, and sends nothing until you say yes.
Works where the work already lives
The problem
Every meeting note ends the same way: a list of things someone now has to go and do. Notetakers write that list well, then hand it straight back to you. The issue doesn't get opened, the follow-up doesn't get sent, and a week later nobody can say who agreed to what.
Navrick carries the task the rest of the way, and it can show you why at every step: what was said, what it did about it, and whether that actually finished the job.

How it works

Records both sides of the call from your own computer. Nobody sees a guest join.

Every task keeps the exact words behind it, who owns it, and what done means.

An agent makes a plan, reads what it needs from your tools, and prepares the action.

You approve the exact send and get a receipt. Then it checks the task really got done. Each meeting also teaches it your people and projects.
One sentence, start to finish
Said at 14:32“I'll open an issue for the export bug today and give it to Priya.”
Done when an issue exists and Priya is assigned.
After your yes: ENG-214 created. Checked, and the task is done.
An illustration of one run. The names and details are examples.
Where it stands
Navrick is in private preview. The whole path runs today on its founder's own meetings and accounts: real meetings, real connections, and real issues, documents, emails and calendar events sent through an approval. The first outside teams come in next, through the list below.
Built by Emin Karka, who spent the year before this building an agent for medical prior authorization, where nothing went out unless a person signed it.
Private preview
We onboard a few at a time, personally, on your real meetings.