Classify
Your categories, applied the same way every time.
Label tickets, documents, emails and transactions into your own categories. The model learns your taxonomy from examples your team has already labeled, including the judgment calls that never made it into a written guide.
Inbox
classify.gguf · on this Mac
Filed
What it does
Built for tasks like these.
- Label every support ticket
- Route inbound texts and emails
- Sort client documents by type
- Flag documents for human review
How it works
From your examples to a model you own.
- 01
Export your labels
Use the categories your team already applies, with real examples of each, including the odd ones.
- 02
Train on your hardware
We train next to the data, on a machine you own. The tax-firm model took 5 days end to end; the bank-intents demo, 30 minutes.
- 03
Label everything
Run it over the backlog and every new item as it arrives. Low-confidence items go to a person.
Private by default
Your documents never leave the building.
The things you classify are usually the most sensitive things you have: client documents, customer messages, financial records. None of them leave the building.
- Your customers' data leaves the building on every call
- The model changes whenever they ship
- You pay for a huge general model to do one narrow task
- The request never leaves the device
- The model is a file you own; it changes when you retrain it
- Small enough to run on hardware you already have
Trained, not prompted
Accurate on your task.
A general model guesses at your categories from their names. Our bank-intents demo learned a bank's 77 support intents from its labelled messages and gets 92% of held-out messages right, ahead of Claude Opus 5.5 at 89%. It is a 30 MB file that runs on a phone's CPU.
Already in the field
A document classifier for a tax firm.
- 5 days
- from raw data to a trained specialist
- 1 MacBook Pro
- on the firm's premises, no rented GPUs
- < 4 hours
- of human time; the rest ran unattended
- 0 bytes
- of client data left the building
And it costs less
No per-call bill.
Classification runs on every item, forever. A model you own makes the ten-thousandth label as cheap as the first.
$120,000 annual savings
See it for yourself
Run this model on your own CPU.
A guided demo, about a minute. It downloads once and runs entirely in your browser; nothing you type is sent anywhere.Open the demoSort and judge
Related solutions.
Have a task like this?
We'll tell you if a micromodel can do it. Or book a demo, or email hello@micromodel.company.