An AI intake assistant for the front desk
Turn an inbound stream of texts and photos into a drafted, bookable work order: OCR on MMS, contact resolution, a job-assembly cockpit, per-org token budgets, and nothing reaching the AI until you send it.
Last week we introduced the Glass module. One piece of it deserves its own post: an AI assistant that works the front desk.
In an auto-glass shop, jobs arrive as a stream of text messages and photos. “Cracked windshield on a 2019 Civic, can you do today?” with a blurry picture of the VIN. Turning that into a booked work order takes a few minutes of typing, and it happens dozens of times a day. Zeno does the first pass now.
From inbound text to a drafted job
An incoming message lands in the intake funnel. Send it to the assistant and it reads the conversation, OCRs the photos including the VIN and damage shots customers send over MMS, resolves the contact against your phone records, and assembles a draft job with the vehicle, the glass part, and the customer. The draft is created early and idempotently, so one conversation never produces two jobs.
Nothing reaches the AI until you send it
No message goes to the assistant until you pick Send to AI or Ignore, so you decide what gets automated. From there the draft lands in the intake cockpit, a job-assembly view where you can inline-edit any field the AI filled in, read the conversation next to the draft, and check related jobs or earlier drafts from the same phone number before creating a duplicate. Booking turns the draft into a work order.
A fixed set of tools
The assistant works through a defined set of tools: check vendor availability, run a GLAXIS inquiry, look up related jobs by phone, and draft the work order. A two-phase loop separates working out what’s being asked from acting on it, so a question about pricing a part doesn’t turn into a booked job. Replies go out signed as the shop, in the shop’s voice.
A token budget per organization
AI that runs up a bill without telling you is a bad surprise. Every organization gets a token budget with usage tracking, so intake automation has a ceiling you set and a meter you can read.
Where the line sits
The assistant transcribes a text thread into a structured job, which is the mechanical part of intake. Which jobs to book, what to quote, and when to push back on a customer stay with your front desk. That’s the same division as the AI Bookkeeper: the assistant drafts, and a person approves before anything becomes a work order.
The intake assistant is live for Glass shops on Zeno. Want to see it on your own message threads? [email protected].