Mission
Listening at the edge of language
We are building tools to explore whether dolphin communication is more structured than we currently understand — and making that research accessible to everyone, not just marine biologists.
The question worth asking
Dolphins are among the most cognitively sophisticated animals on Earth. They have large brains relative to body size, form long-lasting social bonds, teach skills to their young, mourn their dead, and coordinate complex cooperative behaviours. They also produce a rich repertoire of structured sounds — whistles, clicks, burst-pulses — that vary systematically with social context.
The scientific community is clear: we do not know what dolphins are saying. But the evidence that they are saying something structured is compelling enough that it deserves serious, well-resourced investigation.
For forty years the limiting factor has been analysis capacity. Researchers could record more data than they could ever listen to. AI changes that calculus. Models like DolphinGemma can process years of acoustic archives in hours — surfacing patterns no human expert could find manually.
Dolphins Intuition exists to build at that intersection: rigorous science, modern AI tooling, and an open platform that brings this work to people who care about it — not just academic journals.
How we work
Rigour over hype
We will not claim dolphins have human language. We will not exaggerate model capabilities. Every AI output is labelled as interpretation, not translation. The science does not need embellishment — the reality is remarkable enough.
Open findings
Our research logs, model outputs, and audio analysis results are published openly. If we find something interesting, we describe exactly how we found it so others can reproduce, challenge, or extend it.
Tools over papers
We build things: the demo, the analysis pipeline, the dataset tooling. A working tool that lets anyone submit a recording and receive structured analysis is more valuable than a paywalled academic paper most people will never read.
Human-scale accessibility
Cetacean science has a jargon problem. We write for intelligent, curious non-specialists — the same people who read science journalism at its best. Understanding the research should not require a PhD.
Where we are headed
Phase 1 — Live
Complete- →DolphinGemma acoustic analysis demo (Whisper large-v3 + HuggingFace inference)
- →Science explainer: what the research actually says
- →Research log: annotated field notes and AI analysis outputs
- →Open platform on Cloudflare Pages
Phase 2 — In Progress
Current- →Agent-managed content pipeline: AI agent posts research log entries autonomously
- →Mission and contact infrastructure
- →CI/CD: GitHub Actions → auto-deploy on every push
- →HuggingFace model integration improvements
Phase 3 — Planned
Next- →Community audio submissions: field researchers can contribute recordings
- →Comparative analysis: cross-species vocalisation structure
- →Public dataset: curated, annotated dolphin audio with model outputs
- →Integration with Project CETI and Wild Dolphin Project archives
Interested in contributing?
Whether you have field recordings, want to help with analysis, or just want to follow the research — we would like to hear from you.
Get in touch →