Unstructured human data — the layer that decides whether projects live or die.
Infrastructure projects don't usually die on engineering. They die at a podium — in a county board room, at a public hearing, in an ordinance rewrite that happened while the developer was studying soil reports. The data that predicts those outcomes isn't in a GIS layer. It lives in meeting minutes, voting records, local news, social media, and the memory of every project a community has approved or fought before.
Our founders spent their careers building AI products — in political campaigns, national security, and brand management, fields where reading how people form opinions and make decisions is the whole job. When large language models made it possible to read the public record at scale, Navdeep Martin and Eliza Feltus asked where that toolkit would matter most. The answer was infrastructure: critical projects were failing not on technology or financing, but on local social and political risk that no existing tool could see. That's the layer we built Flypower to read — first for renewable energy, then for data centers, where the same patterns play out.
We've been doing this since before siting risk intelligence had a name — reading jurisdictions, scoring social and political risk, and handing developers a verdict: go, no-go, and the path to approval. Today, major renewable energy developers and data center operators use Flypower reports as a standard part of site selection, before capital is committed.
Social and political, fiscal, environmental, zoning — we read every layer above the dirt.
Depth matters here. Reading a jurisdiction — its votes, its meeting minutes, its coalitions — is an AI problem that rewards teams who were building these systems before it was easy to look like one. Every consequential claim in our reports links to the public record behind it.
Two AI veterans take a decade-plus of experience reading how people form opinions — campaigns, national security, brand — to the infrastructure siting problem.
Expanded from renewable energy to data centers, where the same social and political risk patterns were killing projects. One methodology, two verticals.
The two-report system emerges: the Flyover for early go/no-go screening, the Due Diligence Report for sites worth serious capital.
Flypower reports are a standard step in site selection for major developers. Portal, a self-serve search platform, is in development.
Next on the road: Portal — self-serve siting risk search, in development. Illustrative preview.
Two founders, one obsession: the human and political layer of infrastructure.
More than a decade building AI products before founding Flypower. Navdeep leads strategy, methodology, and the standard every report has to meet before it ships.
LinkedIn
A career spent using AI to understand how people form opinions and make decisions. Eliza leads product — how raw public-record signal becomes a verdict a developer can act on.
LinkedInBehind them, a team of analysts and editors who review every finding — and still manage to laugh together about a dozen times a day.
“When ChatGPT launched, I felt something shift. After more than a decade working in AI, I knew this was the moment to build something that mattered. Eliza and I went looking for the biggest problem we could credibly solve — and found critical infrastructure dying in public meetings and local hearings. Not on technology. Not on funding. On the human layer no tool could read.
We started Flypower because the projects that shape our future shouldn't fail for want of understanding the people who decide them. We'd love to show you what's possible.”
— Navdeep Martin Co-founder & CEO
Our AI reads every public signal a jurisdiction produces. Analysts review every finding, and an independent editor reviews every report before it reaches you. We stand by our work.
We serve renewable energy and data centers because social and political risk follows the same patterns in both. That cross-vertical recognition makes our intelligence sharper.
Consequential findings link to the public record behind them — meeting minutes, votes, filings — so you can verify anything we tell you.
Every report ends in a decision: go, no-go, or go-with-conditions — with explicit reasoning and the path to approval, so your team acts instead of interpreting.