The summer that changed everything - and one expat's phone call home
The 2019-20 Australian fire season - the "Black Summer" - burned 46 million acres, killed 33 people, and sent smoke plumes visible from space. Arvind Satyam, a Sydney-born engineer who had spent 10+ years at Cisco building distributed systems, was watching from San Francisco as family back home lived through it. He called Sonia Kastner.
Kastner had spent the previous decade building supply chains for hardware companies at the frontier of AI and connectivity - starting at a thin-film solar startup, then running operations for Whistle Labs (smart dog trackers, later acquired by Mars Petcare), then managing global supply chains for five simultaneous products at Nest. She knew how to source cameras. She knew how to ship sensors. She knew how to build infrastructure at scale.
Together, they asked the only question that mattered: why, in 2020, with ubiquitous compute and AI that could already identify a dog in a photo, could no one catch a wildfire before it became catastrophic? The answer had more to do with distribution and trust than with technology. Fire agencies were fragmented. Data was siloed. And nobody was building the layer in between.
Pano AI was founded in 2020. The name is Portuguese for "cloth" - the same root as panorama. The idea: build a fabric of detection across the landscape, not just a camera in a tower.
We don't move fast and break things - we move urgently, with trust.
- Sonia Kastner, Co-Founder & CEO, Pano AIWhat actually happens when a fire starts
Pano AI mounts ultra-high-definition, 360-degree cameras on the highest ridgelines in fire-prone regions - the same peaks that fire lookout operators occupied before radios. The cameras rotate continuously, and proprietary computer vision models scan each frame for the earliest signs of smoke: faint wisps against horizon, anomalous thermal signatures, the particular light scatter that precedes visible flame.
When the AI flags something, it doesn't just fire off an automated alert. It routes the feed to Pano's 24/7 human intelligence center - trained detection specialists who confirm, triangulate against satellite imagery and sensor data, and push verified, geo-located alerts to the 250+ first responder agencies that subscribe to the platform. The whole cycle: minutes, not hours.
The numbers are stark. In Washington state, the former Commissioner of Public Lands attributed 95% of fires being contained under 10 acres to early detection technology - a benchmark that was impossible before systems like Pano's existed. A 10-acre fire is a ground crew and a few hours. A 100-acre fire is an incident command. A 10,000-acre fire is a national emergency.
The network now spans 10 U.S. states, 5 Australian states, and British Columbia - monitoring forests, grasslands, utility corridors, ski resorts, and private ranch lands. Clients include major utilities like Xcel Energy and Portland General Electric, electric cooperatives, renewable energy developers, and the federal agencies that coordinate wildfire response at scale.
Physics, supply chain, and why Nest was the real training ground
Before Pano AI, Kastner's career reads like a deliberate compression of everything a climate tech hardware company would eventually need. A Harvard physics degree gave her the instinct to follow electrons. A Stanford MBA gave her the framework to follow markets. And a run through some of the most demanding supply chains in consumer hardware gave her the production chops to actually build at scale.
At Nest - the connected home company that Google acquired in 2014 for $3.2 billion - Kastner managed the global supply chain for major electromechanical subassemblies across five simultaneous products, including the early iterations of the Nest Cam. She was inside one of the first mainstream applications of computer vision before most of Silicon Valley had caught on to what deep learning could do to image recognition.
That experience informs how Pano AI talks to its customers. First responder agencies and utilities do not take kindly to false positives. A system that cries wolf loses trust in a matter of weeks, and in the emergency response world, lost trust doesn't come back. Kastner built the human-in-the-loop verification layer into the product architecture from day one - not as a hedge, but as the core value proposition. The AI narrows the search. The humans make the call.
Her stint at NYC's Economic Development Corporation under Mayor Bloomberg - years before she moved to California - added another layer. She knows government procurement, agency politics, and the difference between a press release and an operational partnership. It shows in Pano AI's client roster: real contracts with the agencies that actually fight fires, not pilot programs that never scale.
Time is everything between containing a fire quickly versus letting it get large. Thanks to innovative technologies like Pano AI, the last three years we have kept 95% of our fires below 10 acres.
- Hilary Franz, Former Commissioner of Public Lands, Washington DNR$97 million and the investors who track fire risk for a living
Pano AI's June 2025 Series B - $44 million, led by Giant Ventures - was notable not just for its size but for who else came in. Liberty Mutual Strategic Ventures and Tokio Marine Future Fund are insurance money, and insurance money moves with data. When the largest property insurers in the world decide that an early wildfire detection company is worth backing, it signals something larger than a venture bet: it signals that the insurance industry has priced in the reality that wildfires are no longer tail-risk events.
The earlier cap table - Congruent Ventures, Initialized Capital, Salesforce Ventures - reflects the geography of where climate and enterprise software intersect in San Francisco. Salesforce's presence in particular is telling: Pano AI runs a multi-tenant SaaS platform with enterprise sales, incident management workflow integrations, and a data layer that can plug into existing utility operations tools. It's not just cameras. It's a B2B software business with cameras as the data-acquisition layer.
With 160 employees and $97M in total funding, the company is past the "does this work" phase and firmly in the "how fast can we install cameras" phase. The Series B funds additional coverage, geographic expansion, and the acceleration of Pano's analytics products - including predictive wildfire risk tools that move from detection to prevention.
From Bloomberg-era New York to mountaintop cameras
-
Pre-'07Harvard University - BA in Physics. Then: NYC Economic Development Corporation, Senior Project Manager under Mayor Bloomberg.
-
'07-'09Stanford GSB - MBA. Simultaneously entered California's first solar energy wave.
-
'09-'12Alta Devices - Senior Manager, Supply Chain Development. Built supply chains for a thin-film solar startup at the bleeding edge of photovoltaics.
-
'12-'13Whistle Labs - Director of Operations. Smart IoT dog trackers. Company later acquired by Mars Petcare.
-
'14-'15Nest (Google) - Global Supply Manager. Five simultaneous product lines, including the Nest Cam - one of the first mass-market AI vision products.
-
'16-'17PAX - VP of Supply Chain. Payment and POS hardware at global scale.
-
2020Co-founded Pano AI with Arvind Satyam. Inspired by the 2019-20 Australian Black Summer fires. First cameras deployed within the year.
-
2022-23Awards wave - Entrepreneur 100 Women of Influence; WEF Technology Pioneer; BBC 100 Women; Fast Company Most Innovative Companies (AI). WEF Trillion Trees Challenge Winner.
-
2025Series B $44M closed, TIME 100 Most Influential Companies. WEF Agenda Contributor. 30M+ acres monitored.
The accolades - and what they actually mean
Why an operator, not a researcher, built this
The wildfire detection problem had been worked on by academics, satellite companies, and government labs for decades. What it lacked was someone who could build a hardware-software company from scratch, negotiate contracts with 250 government agencies, design an operations center staffed 24 hours a day, source cameras for remote mountaintop installations, and still run board meetings with Series B investors.
Kastner's background is unusual for a climate tech founder: she's neither a climate scientist nor a machine learning researcher. She's an operator. Her competitive advantage is execution in ambiguous, high-stakes environments where supply chain, trust, and operational reliability matter more than algorithm performance. Pano AI's AI is good - but it's the 24/7 human verification layer and the earned trust of first responder agencies that are actually hard to replicate.
Her time under Bloomberg in New York - before the word "startup" dominated every career conversation - gave her a model for how government-facing businesses have to function: with accountability structures, long contract cycles, and relationships that take years to build. She brought that sensibility to California, where the tech industry typically treats government as an afterthought.
The result is a company that talks to fire chiefs the way fire chiefs want to be talked to. Not with buzzwords about disruption, but with data about detection times, verified alerts, and outcomes that a first responder can put in a year-end report. In a field where false positives destroy credibility, that kind of institutional trust is the product.
We founded Pano AI based on one core belief: that the time to adapt to extreme weather events is now.
- Sonia KastnerFrom detection to prevention - the next frontier
The $44M Series B isn't just for more cameras. Kastner has laid out a roadmap that extends Pano AI beyond detection into predictive analytics - using historical fire data, current sensor readings, wind patterns, fuel moisture indices, and satellite-derived vegetation maps to generate forward-looking fire risk assessments for utilities and land managers. The goal: know where the next fire is likely before it starts, not just catch it faster when it does.
The vision is comprehensive early warning infrastructure for a world where the fire season is now year-round and the WUI - Wildland-Urban Interface, where neighborhoods meet forests - is growing. With insurance companies on the cap table and utilities as paying customers, Pano AI is positioned at the intersection of several enormous markets: utility asset protection, insurance underwriting, government emergency management, and environmental services.
As a World Economic Forum Agenda Contributor, Kastner is also working the policy layer - pushing for regulatory frameworks that incentivize utilities to invest in proactive fire risk management rather than simply reactive response. The argument is economic: a $50,000 camera network installation that prevents a $50 million liability event is a straightforward calculation. The barrier is institutional inertia, and she is working to reduce it one utility contract at a time.
The aspiration is not modest: to make catastrophic wildfires an anomaly rather than an inevitability. Given the trajectory of climate change, the urgency is real. So is the market.