Offline & Local
<10MB model runs on your device. No internet required, no API calls, no latency.
Your machine knows what broke.
It just never told you.
tell is a tiny offline brain you can embed into devices, printers, or code — to explain failures in plain language.
What is tell?
Most devices fail silently. tell makes them speak. It runs offline, interprets internal logs, and gives you answers — not just errors.
Your machine already knows what went wrong. OEM error codes, sensor readings, log patterns — the data is there. tell translates that obscure language into plain English.
Why not cloud LLMs?
Domain-Specific
Trained on OEM manuals and error codes. Speaks your machine's language, not general knowledge.
Purpose-Built
Doesn't need world knowledge to fix your printer. Just needs to understand what broke.
Example Use Cases
3D Printer Diagnostics
Ask your 3D printer what layer failed and get instant troubleshooting advice
Smart Device Monitoring
Ask your smart faucet why water pressure dropped and get actionable solutions
Development Tools
Ask your terminal why the build broke and get clear explanations of errors
How It Works
1
Device sensors or logs
Raw data from your machine's sensors and error logs
2
tell SDK
Preprocesses signals and normalizes data formats
3
Tiny embedded model
Runs completely offline with <10MB footprint
4
Natural-language diagnosis
Clear explanation of the problem and suggested fixes
Documentation
Comprehensive documentation and SDK guides are coming soon. Join the waitlist to be notified when the SDK and docs are available.