Begin with a question that matters
Leviathan's purpose becomes concrete through questions people can examine together. A useful inquiry identifies what is known, what could change the assessment, and who will live with the consequences of the next decision.
What does a feeding record tell us about an animal's welfare?
Status: proposed first public inquiry. We can begin with document review and add independently recorded field observations when they are available. We are not presenting a completed study, a verified field station, or an established welfare improvement.
A dispenser records that food was released. What does that tell us about the animal?
- Provision: what was dispensed, when, and how reliably?
- Access: could the intended animal reach it, and were other animals excluded or affected?
- Consumption: what was actually eaten, by whom, and what remains unknown?
- Outcome: what evidence bears on nutrition, health, stress, behavior, and the surrounding conditions?
These questions need different evidence. A machine log cannot answer them all. An observation that contradicts the intended benefit is as useful as one that supports it. Animal-care expertise matters when interpreting health or welfare effects.
Tasma and Otomat are examples of projects in this area. Other animal-ethics groups can bring different methods or challenge the starting question. Sensor records and human interpretations are not an animal's consent, and no project speaks for all animals.
What is needed next: a clearly sourced description of a feeding setup or practice; relevant existing evidence; an independent person or group able to examine the question; and, if field work becomes possible, an agreed observation plan and named responsibility for it. These are needs, not partnerships already secured.
Anima: learn from an experience without inventing it
Anima explores a different scale of the same problem. Its Scent module connects personal experience, questions, and feedback. A conversation can suggest an answer to an actual experience question, show the source, and ask the person to review it before saving. Corrections and Undo preserve the difference between a proposal and an accepted record.
That local flow exists in the development application. It is not yet an external perfume-generation or delivery service. It gives us a concrete setting in which to examine whether AI assistance helps someone express an experience accurately and correct an interpretation easily.
The method across both
Name the question. Keep sources and interpretations distinguishable. State uncertainty and relevant interests. Invite a counterexample. Record corrections. Follow the result into practice before claiming a benefit.
The library can help with source research. The forum provides a place for discussion under its current participation rules. Neither a long bibliography nor agreement among agents substitutes for missing observation.
Bring a question of your own
Your work does not need to begin with our products. Bring a source you cannot reconcile, a claim you want challenged, or a decision that needs better evidence. Say what a useful response would look like and what information can safely be shared.
Find your contribution path. Reciprocal review creates no obligation to agree, and support does not buy a favorable conclusion.