Backed by Carbon13

Utility intelligence for data centers

We use AI and operational data to detect inefficiencies in cooling, power and compressed air systems, and surface ranked recommendations to operators. No new hardware. Start with chilled water.

Explore capabilities Request a pilot
3–8%
Typical cooling energy reduction
Zero
New hardware required
Days
To first signal, not months
Phase 1
Fully supervised. Operators approve every action
Capabilities

Where we listen

We focus on the utility systems that drive energy waste in data centers: cooling, power and on-site energy assets. We work with your existing operational data and historian signals.

Chilled water cooling plant

Chilled Water Optimisation

Setpoints fixed at commissioning and never revisited. Sono detects headroom and recommends safe upward adjustments, reducing compressor load without thermal risk.

Primary pilot entry point
Data center CRAH units and airflow

CRAH & Airflow

Detect recirculation hot spots, over-cooling conditions and fan inefficiencies across computer room air handlers. Ranked recommendations on supply temperature and airflow balance.

Cooling stack
Power infrastructure and energy monitoring

Power Draw & PUE

Continuous monitoring of IT load versus facility overhead. Identifies UPS inefficiency windows, PDU load imbalance and operational levers to improve PUE.

Power
On-site energy assets

Energy Orchestration

Data centers increasingly operate behind-the-meter assets: BESS, generators, solar and wind. Sono reads signals across these and grid consumption to surface dispatch recommendations, turning sunk capex into active flexibility.

On-site assets
How it works

Three phases of autonomy

We earn the right to act. Sono starts fully supervised and extends autonomy only as it proves value and builds operator trust.

01

Read

We connect to your historian or DCIM and ingest existing signals: chilled water temps, CRAH setpoints, power draw, compressed air pressure. No new sensors. No instrumentation project.

Phase 1 — Supervised
02

Understand

Our ML pipeline detects patterns and anomalies across utility systems. The Utility Reasoning agent applies domain logic to generate hypotheses, ranked by confidence and potential impact.

Phase 1 — Supervised
03

Act

Plain-language recommendations reach the operator with supporting evidence. Operators approve every action in Phase 1. Feedback trains the system. Trust is verified before autonomy is extended.

Phases 1 → 3
Team

Founding team

Three co-founders combining ML, energy infrastructure and commercial execution.

Harit Soni

Harit Soni

Co-founder — Product & Partnerships

15+ years building and scaling cleantech, IoT and AI ventures across Asia, Europe and North America. TEDx speaker and multi-award-winning founder recognized by MIT TR35, Ashden, UNFCCC and WWF.

Johannes Parzonka

Johannes Parzonka

Co-founder — Commercial

Entrepreneurial leader with deep experience in renewables and carbon markets. Founded a green-gas venture and led M&A and business growth across wind, PV and biomethane sectors.

Caterina Giacomelli

Caterina Giacomelli

Co-founder — Technology

Physicist and data scientist specialising in ML, signal processing and anomaly detection. Experienced in energy infrastructure and predictive modelling for data center operations.

Contact

Work with us on an early pilot

We are working with data center operators on early pilots. If you are running a facility and curious about what your historian data could tell you, reach out.

harit@sono.so · johannes@sono.so · caterina@sono.so