Applications

What it is built for

A narrow specialist you fine-tune, then ship inside the firmware.

Atome LM is not a general-purpose assistant. It is a narrow specialist you fine-tune on the data your product actually sees, and then it ships inside the firmware. That constraint is what makes it useful: a model that only has to understand one vocabulary can be small enough to live in flash, and a model that lives in flash never sends a customer's voice anywhere.

Where it fits today

ProductThe jobWhy on-device
Smart lightingLocal voice and text commandsWorks when the Wi-Fi does not
Children's toys and dollsOffline interactionNo recording ever leaves the device — the strongest possible answer to a parent's question
Bedtime story devicesOffline generationNo subscription, no outage, no account
Pet feeders, litter boxesFriendly status messagesA $2 part cannot afford a cloud round trip per event
AutomotiveVoice-intent detectionLatency and certification both rule out the network
Watches and wearablesText comprehension at the wristBattery: a radio costs far more than an inference
AgricultureField-sensor pattern recognitionThere is no coverage in the field
Medical wearablesECG and vitals classificationData residency, and a device that must keep working offline
Industrial sensorsAnomaly detectionAir-gapped plants, and per-decision cost at high sample rates
Energy and utility metersOn-meter reading parserRadio payload drops by roughly 48× when the meter sends a decision instead of a waveform
Hearing aidsOn-device sentence completionMicrowatts, and no microphone stream leaving the ear
Disaster-relief radiosField text help, off-gridThe network is the thing that failed

Three prototypes that exist and run

Trained, exported, and run through the Cortex-M3 emulator. The engine path for running them is public; the training scripts for these specific tasks are not.

PrototypeWhat it doesDataHeld-out accuracy
Wake-word / command intentPicks the right command from text variants6 classes, 1,800 synthetic samples with lexical variation100%
Anomaly flagSpots bad sensor-reading strings — NaN, out-of-range, garbage spikesBinary, 1,000 samples, 30 epochs91.7%
Intent bucketSorts a sentence into command / question / status / alert / greeting5 classes, 1,500 samples, 40 epochs100%

Read those two 100% figures as what they are: synthetic distributions, generated with simple lexical variation. A saturated accuracy on generated data is a sign the task is easy, not a sign the model is remarkable. They demonstrate that the engine shape supports a classification head and that the export path works end to end. They are not evidence about your data.

v2 — twelve apps on a $5 board

Atome LM v2, "SuperESP", extends the ternary engine from text generation into 12 on-device applications: agriculture, voice, motion, anomaly, air quality, energy and non-intrusive load monitoring, occupancy, wearable, water leak, predictive failure, sound, plus an on-device dispatcher. It ships a universal installer that flashes any ESP32 variant — S2, S3, C3, C6, H2, Xtensa or RISC-V — a bring-your-own-CSV training path, Ed25519-signed models with a load-time integrity check, and a tamper-evident decision log.

Honest scope, in the same breath: several demo heads ship on physics-grounded synthetic data, clearly labelled and swappable with your own. Voice keyword spotting works but is modest on a 20 KB ternary model. Only the ESP32-WROOM-32 is silicon-tested; the other variants are build- and QEMU-verified. The v2 write-up has the detail.

Three shapes of task that fit

If your product is in the table and you want to know whether the model is small enough for the part you have already chosen, describe the part and the task — that question has a measured answer rather than an opinion.

Questions

What can you actually build with a 944K-parameter language model?

Narrow things, done well: a product FAQ that answers offline, command and intent classification, sensor-string anomaly flagging, status message generation, keyword routing. Trained on one domain it is fluent inside that domain. Trained wide at this size it produces incoherent text.

Is Atome LM suitable for a children's toy?

It is one of the clearest fits, because the model runs entirely inside the firmware with no allocator, no sockets and no network. No audio or text ever leaves the device, which is a property of the compiled binary rather than a policy commitment.

Can it run on battery?

The footprint says yes and the measurement does not exist yet. No Joules-per-token figure has been taken on real hardware, so anyone quoting one for Atome LM is quoting something we did not measure.