Editions
The same mind, in builds that differ in which models run its parts and where it runs. Pick one and talk to it. You can switch at any time.
Standard edition
Each part runs on a model from a different AI lab: OpenAI, Anthropic, xAI, Meta, and Perplexity. The voice runs on Anthropic's Claude. The fullest version of the mind, run by Sheaf.
Talk to the standard editionOther country editions
The same mind built on a country's own AI model, so it thinks in that country's language and can run completely inside that country, on servers the institution controls. No question and no memory has to leave the country.
Qatar edition: every part on Fanar, the model built by Qatar Computing Research Institute. Arabic or English. About it.
Spain edition: every part on ALIA, the model built by the Barcelona Supercomputing Center for Spain's public AI programme. Castilian, Andalusian, Canarian and other varieties, plus Catalan, Valencian, Basque and Galician. About it.
Built for the questions that are about being human
A conversation you are dreading. A parent who is changing. Whether to take the job, leave the city, or end the thing. What to say to someone who is grieving. Whether you are the problem. What you actually want.
These questions have more than one side, and the sides pull against each other. A single model has one temperament and no way to notice when its own answer is pulling in two directions, so it flattens the question into a list. Sheaf Mind sees the whole situation, names what is really going on, and gives you a read.
The questions it is not for
Facts, research, code, spreadsheets, and "summarise this document". A single model with web browsing does these as well or better, and faster. Use it for those. Sheaf Mind is for the questions where the answer depends on people.
What is inside
Each part is a separate job, run by a model from a different AI lab, so the parts do not share one blind spot. They all read your question at once. Then the voice reads what every part found and answers.
Emotional reasoning
Reads how the people involved feel and what they need to hear.
Conscience
Pushes back. Says what is unsure, what is being avoided, and what would be a mistake.
Reason
Works out what is true, what follows, and what the trade-offs are.
Imagination
Finds the options and the reframes you had not thought of.
Memory
Keeps a few things it learns about you, so the next conversation starts where the last one ended.
Perception
Checks live facts on the web when the question depends on today, and cites them.
The check
Before the voice speaks, Sheaf's coherence engine measures whether the parts agree. If they do not, the answer says so and names the tension.
The voice
One part speaks for all of them. It commits to a read and ends a decision with a clear call.
The parts run on models from OpenAI, Anthropic, xAI, Meta, and Perplexity. The voice runs on Anthropic's Claude. You can watch every part's notes as it thinks. Nothing is hidden.
How it compares
| A single model | Sheaf Mind |
|---|---|
| Gives you a list of options. | Tells you what it thinks, and why. |
| Answers the question you typed. | Answers the question underneath it. |
| One model reads your question and answers. | Five parts read it at once. One voice answers for all of them. |
| Its own doubts stay inside the model. You cannot see them. | Sheaf measures whether the parts agree. Disagreement is named in the answer. |
| Confident on everything, including what it does not know. | Conscience has a seat. The answer says when it is unsure. |
| Starts from zero every time. | Remembers you, so the next conversation starts where the last one ended. |
The blind test
Ask a question. You get two answers with no names on them, one from Sheaf Mind and one from ChatGPT. Pick the better one. Then we show you which was which, and how long each answer was.
What it is like to use
You talk to it the way you talk to any AI assistant. Under each answer you can open the reasoning and see what Reason argued, what Conscience objected to, what Imagination offered, and how the voice weighed them. Decisions end with a line that states the mind's read. When the parts pull apart, the answer says so in one sentence and tells you what would settle it.
The first 20 messages are free and need no account. After that it is $5 for 50 messages that never expire, or $12 a month for 300 messages and memory that follows you across devices. It remembers you on this device either way.
Data handling
Sheaf Mind keeps a small number of facts it learns about you, tied to this device, so it can remember you. Nothing is used to train any model. Each AI model provider that runs a part of the mind receives your question and is named as a sub-processor. You can ask the mind what it remembers about you.
The mathematics behind it
Sheaf Mind uses sheaf cohomology. Each part of the mind is one local piece, and before the voice speaks the mathematics asks whether all the parts fit into one whole. A tension that only shows up when the parts are put together is found and named in the answer. When the parts disagree, anything the mind was asked to do on the strength of that answer waits for a person.
Ask it the thing you have not said out loud.
Talk to Sheaf MindPut it up against ChatGPT, blind20 free messages, no account needed. Then $5 for 50 or $12 a month for 300. Or get in touch.