Vaticore Engine

Tomorrow's grid, planned today.

Vaticore forecasts demand, solar and the grid for every site, then tells your team when to run the generator and when to let the battery carry the load. Less diesel, fewer blackouts.

Vaticore/Sites/Solar mini-grid, 120 kW
Live replay Tue 17 Mar, 00:00

Net load, yesterday and next 24 hours

Demand minus solar, % of peak. Forecast issued at midnight.

Real Elia grid data, replayed hour by hour for a representative solar mini-grid. Hover the chart for values.

47%

lower forecast error than persistence, level with a grid operator's own forecast

80.4%

of outcomes inside the 80% range. Calibrated, not guessed

19.6%

less diesel at an off-grid tower than the best-run practice, over a held-out year

0

control actions. Advisory only, a person always decides

Accuracy

Measured, not promised.

Every model is scored on held-out history against persistence, the forecast that assumes tomorrow looks like today, and, where one is published, against a grid operator's own forecast. If a model can't beat persistence, it doesn't ship.

Pinball loss

Rewards a range that is both tight and honest. Lower is better.

−47%
Persistence
178.4 MW
Elia (grid operator)
95.2 MW
Vaticore
95.2 MW

Mean absolute error

How far the expected forecast lands from what happened.

−44%
Persistence
510.0 MW
Elia (grid operator)
292.4 MW
Vaticore
287.8 MW

453 held-out days of Belgian national load. Using calendar and past load only, no weather, Vaticore matched the day-ahead forecast Elia publishes, and its range was the better calibrated: 80.4% of outcomes inside the 80% band, against Elia's 74.9%. On solar, where tomorrow depends on the weather, Elia's weather-driven forecast is still far ahead. Weather is our next study.

What it's worth

Less diesel, from the same generator.

We replayed a full year the planner had never seen, against how sites run their generators today. Every day ran on what actually happened, and we counted the litres.

−82.6%

diesel at a tower on a weak grid, against starting the generator every time the grid fails

−19.6%

diesel at an off-grid tower, against the best-run practice, with no outages

−30.4%

diesel at the same tower, against an evening generator timer

−68%

outage hours at a solar mini-grid (94 to 30 a year), on 2% less diesel

Most of the saving comes from how the generator is run: let the battery go first, and when the generator runs, run it hard, then off. Vaticore finds the right settings for each site from its own history, plans every day, and scores every plan against what happened.

Climate

Every litre not burned is 2.66 kg of CO2 kept out of the air.

Diesel generators back up millions of sites where the grid fails. Vaticore cuts the litres, and measures them, so the carbon saved is counted, not guessed.

4 t

CO2 a year less at the off-grid test tower, from 19.6% less diesel

14 t

CO2 a year less at the weak-grid test tower, by letting the battery go first

255,000 t

CO2 a year if Nigeria's telecom towers cut their diesel by a fifth. A scenario, not a forecast

The same daily measurement that proves a saving to an operator can back its carbon reporting, and the carbon finance that pays for solar and batteries.

Why Vaticore

Built for the grid you actually run.

A range, not a guess.

A battery sized for the average day fails on the worst one. Every forecast carries its likely range, so you plan for the top of it.

Answers in plain words.

No chart reading required. The plan arrives as one clear call your team can act on.

Today's planSolar mini-grid

Hold 600 kWh tonight. Evening demand runs past what the battery can cover, so plan the generator for the peak.

Made for messy meters.

Real feeds drop readings. Vaticore detects every gap, logs it and handles it explicitly. Never silently.

3 missing readings flagged and handled

It advises. You decide.

Vaticore never touches dispatch. Every recommendation goes to a person first. That's a line we don't cross.

Forecast and advise
Control your equipment

An API your team can build on.

Every operator and site is scoped from the first call. The same engine drives the dashboard.

POST /advisory { "site_id": "ikeja-minigrid", "usable_battery_kwh": 600 } // 200 OK { "recommended_reserve_kwh": 600.0, "genset_recommended": true, "expected_net_load_kwh": 1762.2 }

How it works

From meter data to a decision, every hour.

01

Connect

Send a CSV or stream readings over the API. Any resolution, any timezone, gaps expected. Everything is validated on the way in.

02

Forecast

A quantile model learns each site's demand and solar from calendar, recent history and weather, and sharpens as more data arrives.

03

Plan

The forecast becomes a reserve and backup recommendation with its range, refreshed as each new reading lands.

Who it's for

For operators who can't afford a surprise.

C&I solar and storage

Hold the right reserve before the clouds arrive.

Forecast building demand against on-site solar and stop being surprised by the afternoon.

Batteries and flexibility

A sharper net load is a better dispatch plan.

Calibrated demand and generation ranges feed straight into your optimiser and market bids.

Solar mini-grids

Run the generator on purpose.

Know tonight's shortfall this morning. Fewer litres burned, fewer outages for the people you serve.

Run a pilot on one of your sites.

Send six to twelve months of meter history. We forecast it, score it against your own baseline and show you the plan. Free, and your data stays yours.