Open source
Jev: sort 2,000 support tickets for 4 cents, not $10
A support inbox gets two thousand tickets every morning. Jev, an AI model from TypeSafe, answers three questions about each ticket in one request (what the customer needs, whether it is urgent, how upset the customer is), with confidence scores, and your code routes it: billing, an automatic reply, or a person. Unsure tickets go to a person. Reading the same 2,000 tickets costs about $9.46 in input with a top model from OpenAI or Anthropic, and about $0.04 with Jev. Price from typesafe.ai, checked Sep 21, 2026; tokens per ticket from one live Jev run. Tickets and names in the demo are fictional.
Watch on YouTube · Watch on TikTok
The story
What the video said
An AI model can read two thousand support tickets, sort them, and hand the urgent ones to a person, for about four cents. Here is how that works in your inbox.
Right now your team reads every ticket just to decide who should answer it. Urgent problems wait in line, and customers who wait too long cancel and take their money elsewhere.
This is Jev, an AI model that reads each ticket and answers questions about it. Jev comes from TypeSafe, a company founded by former OpenAI researcher Diogo Almeida.
For every ticket, Jev answers three questions at once. What does the customer need? Is the ticket urgent? How upset is the customer?
Your code turns those answers into a route. Billing problems go to the billing team. Password questions get an automatic reply. Urgent tickets go straight to a person.
When Jev is not sure, the confidence score says so. This ticket could be billing or a bug, so your code sends it to a person instead of guessing.
Sending those tickets to a top model from OpenAI or Anthropic costs about ten dollars, for the input alone. Jev reads the same tickets for about four cents. Would you let Jev sort your support inbox?
This page was generated from the Ming Dao AI publication record.
Ming Dao AI is independent and unaffiliated with GitHub or the featured project. Read our editorial policy.