Our Stance on AI
Focused AI, pointed at a real problem.
Pelagos is a conservation and environmental company that builds software. That ordering shapes everything below: sustainability is the first question we ask about any use of AI, not the last one. Within that constraint, we use AI in controlled environments, in select and focused processes, where it makes the work dramatically more efficient. Fisheries management runs on evidence, expert judgment, and public trust, and none of that should be handed to a black box. So we want to be clear about how we use AI, how we do not, and what we require of the infrastructure it runs on.
We use AI differently
Built to be right about one thing.
Most AI you meet is built to sound convincing about anything. The Fisheries Intelligence Platform is built to be right about one thing: the U.S. fisheries record. It works over a curated, authoritative body of council decisions, stock assessments, regulations, and the science behind them, and it is engineered to ground its answers in those sources and cite them. It is a research instrument for a specific field, not a chatbot with opinions.
AI serves the experts, it does not replace them
The people stay in charge.
The people who manage fisheries stay in charge: the scientists, the council and agency staff, the attorneys, and the fishermen who live with these decisions. Pelagos does the work machines are genuinely good at, which is reading everything, cross-referencing it, and surfacing what matters, so people can spend their time on judgment instead of retrieval. Every answer traces back to its source, so it can be checked rather than trusted on faith.
On energy, water, and the environment
Deliberately, efficiently, and only where it counts.
We will not pretend AI is free of environmental cost. Training and running large models uses real energy and water, and a company that works to protect the ocean has no business ignoring that. Wise, minimal use of resources is itself a conservation ethic, and we hold our own operations to that standard, not just the fisheries we study. So we use AI the way we believe fisheries should be managed, which is deliberately, efficiently, and only where the benefit clearly justifies it.
Pelagos runs a focused tool over a curated body of knowledge. We are not building ever-larger models, and we do not spend compute for its own sake. Pointed at the right problem, this kind of AI returns far more than it costs. In fisheries, the price of leaving the record scattered is paid in collapsed stocks, wasted public effort, and decisions made in the dark.
Data centers and conservation
Where the servers sit matters as much as how they run.
The physical footprint of AI is real, and it belongs somewhere specific. Data centers are industrial facilities, and they should be sited on land already zoned for industrial use, not carved into dense residential neighborhoods or built in places already under strain. A dense, coastal county like Pinellas, with limited land, a stressed water supply, and a large residential population nearby, is the wrong place for this kind of build. Siting is a land use and utility decision as much as a technology one, and it deserves the same scrutiny we bring to any resource management question.
Where a facility sits also determines its environmental cost. The regional grid mix decides whether the power behind a query comes mostly from hydro and wind or from gas and coal. The cooling method decides how much water a facility pulls from a local watershed, since evaporative cooling can draw heavily on supplies that communities and ecosystems already depend on. We weigh grid mix, cooling method, and regional water use when we choose where to run our own infrastructure. Pelagos thinks through the physical locations where its data and its processing are conducted, so that they are the most environmentally friendly and efficient they can be.
Why it matters
Information as good as the decisions deserve.
For decades, some of the most consequential natural-resource decisions in the country have been made with tools from another era, with people copying figures into notebooks and searching a dozen portals to answer a single question. Bringing careful, modern AI to that work is not a threat to fisheries. It is one of the most useful things it can do, because it gives the people protecting these resources information as good as the decisions deserve.
See for yourself
We would rather earn trust than assume it.
If you want to see exactly how this works, look at the platform.