Edviro, a Y Combinator-backed startup, has deployed artificial intelligence that autonomously operates heating and cooling systems across 34 school facilities, saving what the San Francisco company reports as $400,000 in energy costs.
The software connects directly to building equipment and detects waste in real time, making adjustments without requiring districts to rip out existing infrastructure. According to Edviro's launch materials, the system ties into utility meters, HVAC controls, building automation platforms, and work order systems.
The technology arrives at a moment when school districts are hunting for budget relief wherever they can find it. What makes Edviro's approach distinctive is the level of autonomy it claims: the AI doesn't just flag problems for human review but simulates interventions, tests changes, and then implements them.
Here's how it works. The platform ingests data from utility bills, sensors, and management systems to construct what the company calls a "living digital twin" of each building. Before making any adjustment, the system runs fault detection and simulates the outcome. Only after verification against actual meter data does it proceed.
Temperature setpoints get dialed up or down. Equipment schedules shift. Operators retain veto power over what Edviro deems "consequential actions," though the company hasn't defined where that threshold lies.
One feature drew my attention for its ingenuity: Edviro piggybacks on existing Wi-Fi routers to track whether anyone is actually in the building, then adjusts heating, cooling, and lighting accordingly. No new sensors required.
Early Traction in Silicon Valley's Backyard
Edviro is live with Palo Alto Unified School District and Los Gatos-Saratoga Union High School District, according to logos on the company's homepage. Both are Bay Area districts, which raises the familiar question of whether a solution will scale beyond early-adopter communities with above-average technical sophistication.
A blog post from late July described how the software's portfolio view across seven schools surfaced an anomaly: one site burning roughly five times the energy of comparable facilities. The platform can surface and route work orders with guided workflows, generating what Edviro describes as "board-ready, audit-grade" measurement and verification reports from a learned baseline.
The founders are strikingly young. Hursh Shah and Tanuj Siripurapu, both 19, started the company after Shah uncovered what he calculated as $360,000 in gas overbilling while analyzing utility data for a high school district. Shah has contributed research to a NeurIPS workshop and previously founded Skyglass, which was accepted to PearX. Siripurapu worked at RTX and built Edviro's iOS app, which hit the App Store in late May.
"The long-term goal is not another dashboard," the pair wrote in their Y Combinator launch post. "It is infrastructure that can understand itself and increasingly operate itself."
Bigger Game: Data Centers
Schools may be the proving ground, but Edviro has larger ambitions. The company is developing data-center-specific models and integrations, with a focus on the tightly coupled constraints of power and cooling in those environments. For now, it plans to operate in read-only mode in data centers, requiring operator sign-off on any changes.
The timing aligns with a surge in electricity demand driven by AI infrastructure. Data centers are projected to account for nearly half of U.S. electricity demand growth through 2030, per recent International Energy Agency estimates. The Energy Information Administration has called it the strongest four-year demand spike since 2000.
Edviro raised funding from Y Combinator and Reach Capital, though the company declined to specify amounts. Shah announced the Reach Capital investment on LinkedIn in July. The company is part of Y Combinator's Summer 2026 batch.
A Crowded Field
The competitive landscape is heating up. BrainBox AI, which Trane Technologies acquired in January, operates in similar territory. Phaidra focuses on what it calls "agentic AI" for data center cooling and has disclosed testing partnerships with NVIDIA. Verdigris offers electrical intelligence paired with submetering hardware, while GridPoint targets multi-site commercial and retail customers.
Edviro is offering free energy audits to early district partners, according to a LinkedIn post tied to the launch. Pricing scales with portfolio size and the number of connected sites, but requires a demo to get a quote.
Whether autonomous building management catches on broadly will depend on trust as much as technology. Facilities managers at schools and data centers alike tend to be conservative, for good reason. Equipment failures can mean frozen pipes in January or server outages that cost thousands of dollars per minute. Edviro's approach of simulating changes before implementing them may help, but the company will need to prove reliability across a wider range of environments and edge cases.
For now, it's two teenagers with a Y Combinator pedigree and a handful of Bay Area school districts willing to let an algorithm take the controls.
