Edviro: AI that operates energy infrastructure

Hi all,

We’re Hursh and Tanuj, founders of Edviro.

TL;DR: Edviro connects to a facility’s BMS, utility, sensor, equipment, and maintenance systems to build a world model of how the facility behaves. We use that model to detect problems, simulate interventions before anything is changed, and coordinate the best next action. We’re live in 34 facilities, have identified more than $400K in avoidable energy costs, and are now opening our first partnerships with data center operators.

https://youtu.be/0sTQlaQGlOY

The problem

Data centers are expected to drive roughly half of U.S. electricity-demand growth through 2030.

This is usually framed as a problem of producing more electricity. That is only part of it.

Power must be delivered to the right sites, on the timelines data centers are being built, with the reliability their workloads require. Once a facility has that power, its operators still need to extract as much useful compute as possible from every available megawatt without compromising uptime.

Inside a data center, power, cooling, controls, equipment, workloads, and maintenance are tightly coupled. A workload shift changes the thermal profile. A cooling-control change affects both energy use and equipment margins. A maintenance decision can alter redundancy and available capacity.

Yet the data and workflows used to manage these systems remain fragmented across BMS, DCIM, EPMS, meters, sensors, maintenance software, vendor portals, and spreadsheets.

Most existing systems can display telemetry or raise an alarm. They cannot reliably model how the entire facility will respond before an operator changes a setpoint, shifts a workload, services equipment, or adds capacity.

Operators are left making high-stakes decisions with incomplete system-level context.

What Edviro does

Edviro turns a facility’s operational data into a continuously updated world model.

By “world model,” we mean a model that estimates the current state of the physical system, learns how its components interact, and predicts how it will respond to a proposed action. 

Edviro:

• Integrates data from building controls, utilities, sensors, equipment, and maintenance systems
• Detects anomalies, energy waste, equipment degradation, and capacity constraints
• Simulates operational and maintenance interventions before they are deployed
• Recommends the best action while accounting for cost, reliability, and physical constraints
• Uses agents to coordinate approvals, work orders, investigations, and follow-through
• Verifies whether the expected operational and energy impact actually occurred

For a data center, this could mean determining whether existing power and cooling infrastructure can support additional IT load, testing a control-sequence change before deploying it, identifying which equipment issue is constraining capacity, or prioritizing maintenance based on the capacity and reliability it would recover.

We initially keep operators in control of every consequential action. As the models are validated and earn trust, more of the operating loop can be automated.

Why we started with schools

Edviro began when I was 17 and working with my high school’s facilities team. I was given a year of district utility bills to analyze, built software to parse them, and discovered roughly $360,000 in gas overbilling.

We found that this was not an isolated problem. Schools operate large portfolios of aging, mixed-vendor infrastructure with fragmented controls, incomplete data, lean facilities teams, and very little room for wasted budget.

They forced us to build for the messy reality of physical infrastructure rather than clean laboratory data.

Today, Edviro is live in 34 facilities and has identified more than $400,000 in avoidable energy costs for building operators.

A school is obviously not a data center. Data centers have denser instrumentation, much tighter uptime requirements, and more complex coupling between power, cooling, and compute. 

What transfers is the underlying architecture: connecting heterogeneous operational systems, estimating the state of a physical facility, simulating interventions, coordinating action across teams, and verifying the result. We are now building the data-center-specific models, integrations, and safety constraints directly with operators. Our first relevant research benchmark can be found here: https://x.com/hursheybar2/status/2083299315003593037?s=20

Where we’re going

The current generation of facilities software tells operators what has already happened.

We want Edviro to understand what is happening now, predict what will happen next, and help execute the safest and highest-value response.

As electricity becomes one of the primary constraints on AI growth, building more infrastructure will not be enough. We also need to operate the infrastructure we already have far more intelligently.

The long-term goal is not another dashboard. It is infrastructure that can understand itself and increasingly operate itself.

Asks

1. We’re looking for data center operators to become early design partners. We are especially interested in facilities dealing with power or cooling constraints, capacity-expansion planning, recurring manual investigations, fragmented controls, or reactive maintenance. We can begin read-only, with operators retaining approval over every action.
2. We’re hiring energy-modeling engineers across thermal systems, power systems, controls, scientific machine learning, and physical simulation.

If you operate a data center—or know the team responsible for operating one—we’d love to talk.

www.edviro.com