Offline-safe edge
A stale cloud schedule triggers a local fallback mode, peak shaving or balancing, or a safe hold. Telemetry is buffered and flushed on reconnect.
Myto / Platform
Your hardware. Our intelligence.
An edge controller on site, forecasting and optimisation in the cloud, and one model of the energy system joining them. Vendor-neutral below, auditable above.
Edge + Cloud
Control where the plant is. Thinking where the compute is.
The edge runs a one-second read–decide–write cycle and never depends on the cloud to stay safe. The cloud forecasts, optimises and sends schedules down. If the link drops, the edge falls back to a local mode and buffers telemetry until it returns.
Northbound to SCADA
A Modbus TCP server and an IEC 60870-5-104 outstation expose the site to an existing SCADA or an SLDC-style telemetry link.
Deploy where you need it
Containerised services on your own infrastructure or ours. Where the data lives is a deployment decision you make.
Integrations
The protocols we support, and the ones we do not yet.
Vendor-neutral means nothing without a list. This one is kept to what the edge actually speaks today.
AvailableBy design
InterfaceRoleStatus
Modbus TCPSouthbound client for inverters, BMS, PCS, meters and chargers, with device profiles.Available
Modbus TCP serverNorthbound holding registers for SCADA.Available
MQTTTelemetry publishing and setpoint subscription.Available
OPC UASouthbound client, configured per site.Available
BACnet/IPBuilding systems, configured per site.Available
IEC 60870-5-104Master for PCS and RTUs (core subset), and an outstation for SLDC-style telemetry.Available
DNP3Master over TCP (core subset).Available
OpenADR 2.0bVirtual end node for demand-response signals.Available
Weather forecastsWeather feed used by generation forecasting at the edge.Available
IEC 61850Lab mapping of logical-node paths to channels. Not a certified MMS or GOOSE stack.By design
REST APIHTTP API for the platform; an OpenAPI description covers the market and optimisation module.By design
OCPPCharge-point management protocol for EV chargers.By design
M-Bus · KNXMetering and building-automation buses.By design
Prediction
Predict. Detect. Recommend. Learn.
We use the word AI sparingly. Where a model makes a decision we say what kind of model it is, how it was tested, and whether it ships today.
Load, generation and price
An ensemble price forecast in the cloud with Indian calendar effects; persistence and similar-day predictors for load and generation at the edge.
AvailableFaults and drift
Rule-based fault detection on live channels, raised as alarms. Learned anomaly detection is by design.
AvailableSchedules with reasons
Storage schedules and bid drafts produced in shadow mode, for an operator to review before anything is armed.
By designMeasured, then promoted
Forecast models are backtested on held-out history, promoted when they measure better, and rolled back when they do not.
AvailableGradient-boosted and neural forecasting models have an interface in the platform but are not in production. We publish no forecast accuracy figures until they are measured on a deployment.
Optimisation
A schedule that knows what a cycle costs.
The optimiser turns forecasts, tariffs, prices and constraints into a storage schedule, and prices every cycle in degradation before committing to it.
Degradation-aware scheduling
A per-kWh degradation cost from the asset profile sits inside the objective, so a cycle is only spent when it earns more than it wears.
Reservation price
A floor below which the battery holds its energy rather than selling it cheaply.
Flexibility horizon
A horizon schedule for flexible load and storage, activated to the edge as a schedule to follow.
Solver choice
A built-in dynamic-programming solver by default, with an optional mathematical-programming solver for larger problems.
Optimiser output runs in shadow mode. It becomes a physical setpoint only after live control is armed at the edge.
Digital twin
Simulate before you invest.
A storage asset modelled at 15-minute resolution, run against real history before a rupee is committed.
You provide
- Load and generation history
- Tariff or market exposure
- Battery power, energy, efficiency and degradation cost
- Connection and contract limits
You get back
- The schedule the optimiser would have run
- State of charge and cycles, block by block
- Indicative revenue, payback and NPV
- Where the value came from, by stream
BESS digital twinState of charge, power and degradation at 15-minute resolution.By design
EconomicsIndicative profit, payback, NPV and IRR. Estimates, not a bankable financial model.By design
Engineering sizingEnergy, power, PCS and thermal sizing for design.By design
Technical credibility
How it is built.
For the engineer who will be asked to sign off on it.
Stack
Go services for the API, edge agent and workers. PostgreSQL as the system of record. A web operator console. Containerised deployment.
Control path
One-second input, process and output cycle at the edge. Cloud schedules are followed only inside the local envelope, through the dispatch gate.
Evidence trail
Control and market actions written to audit tables. Compliance runs checksummed. Backtests replayed without look-ahead leakage.
Simulated equipment
Meter, storage, charger and thermal simulators, so control logic is exercised before it meets real plant.
Access
Session and API-key authentication, organisation-level roles, and separate privileges for market functions.
Origin
Developed in India. No dependence on a foreign closed-source control plane.
Want the technical briefing?
Architecture, protocol mapping for your equipment, and the control modes we would start in. For engineers, not for a sales deck.