Mathematical intelligence infrastructure.
How Panbiota turns complex data into intelligence an AI system can use, and where each part stands today.
From data to the AI that uses it.
- 01Data
Structured, continuously updated information from each field, every figure with its source and time.
Live - 02Mathematical discovery
Methods that find relationships in the data and keep the useful ones as reusable mathematical structures.
In development - 03Verification
Exact, repeatable calculation, provenance on every figure, and outputs that can be checked independently.
Live · provenanceDemonstrated · exact checking - 04Intelligence API
Controlled access to Panbiota's capabilities for AI systems and software.
Live · AI access pointIn development · keyed API - 05AI models and applications
Frontier AI models, autonomous agents, analytical platforms and enterprise systems.
Yours
What each layer does, and what is live.
Data
Structured, continuously updated domain information. Panbiota reads public sources field by field, turns each into a consistent structure, and keeps every reading with its source, its licence and the time it was read.
- Live today
- 48 live instruments across 41 fields, from power grids, shipping and sanctions to vulnerabilities, recalls, filings and markets, and the record of every reading since 17 October 2026.
- In development
- Further domains, each built and kept by Panbiota and answered through the API.
Mathematical discovery
Methods for discovering useful relationships in data and keeping them as reusable mathematical structures. A relationship found once and written exactly can be checked, combined with others and used again.
- Live today
- Not yet part of any public service.
- In development
- Discovery methods, applied first to Solana, where every relationship found can be checked against what followed. A relationship is kept only if it holds beyond the data it was found on.
Verification
Deterministic calculations, provenance and independently checkable outputs. A result should be reproducible from its inputs, traceable to its sources and checkable by someone else.
- Live today
- Every figure carries its source, its licence and its time. The daily notes are computed from the recorded figures by fixed rules: the same inputs give the same line.
- Demonstrated in research
- An exact core that checks whether a set of statements is consistent, works out what follows from them with a full calculation trace, and refuses rather than guesses when an answer cannot be determined. It runs inside our research, not yet as a public service.
- In development
- Verification and calculation as services through the API.
Intelligence API
Controlled access to Panbiota's capabilities, for AI agents and for software. The API sells answers, not the underlying assets: the engine, the relationships, the history and the discovery behind each answer stay private to Panbiota.
- Live today
- An open access point for AI assistants and agents at panbiota.ai/mcp, on the Model Context Protocol, the open standard AI systems use to connect to tools. It serves the latest readings and daily notes of 43 instruments whose sources allow reuse, each with its credit.
- In development
- Secure, paid access for AI agents and software: accounts and keys, pay per use with prepaid credits and spending limits, audit logs and isolation between customers.
AI models and applications
Frontier AI models, autonomous agents, analytical platforms and enterprise systems sit on top. They do the reasoning and the conversation; Panbiota supplies the intelligence they reason over.
- How they fit
- A model may propose a question or a hypothesis; it cannot certify a Panbiota result. What Panbiota computed is always kept apart from what a language model wrote.
- Independent of any model
- Panbiota does not build language models and is tied to none. Where Panbiota's own services use a third-party language model, we say so.
Compute once. Verify continuously. Serve many times.
Panbiota is designed to do the work before anyone asks. For its first domain, Solana, the system would continuously:
- Ingest relevant Solana blockchain events
- Identify wallets, creators, funding relationships and transaction patterns
- Apply mathematical discovery and verification
- Maintain historical relationships and their evidence
- Update stored intelligence when new information arrives
- Keep verified intelligence ready to retrieve
When an AI agent asks, Panbiota retrieves intelligence already computed wherever it can; a complex new question may still need fresh computation. The 48 public instruments already work this way at small scale: each is read continuously, computed once, and served to every visitor and AI assistant from what is kept.
Three operational layers.
Ingestion
Reliable, continuous collection of blockchain information, with gaps detected and filled and every event kept in order.
Intelligence
Mathematical discovery, verification, relationship indexing and historical knowledge, updated incrementally as events arrive rather than rebuilt from scratch.
Delivery
Secure, low-latency APIs that serve specific answers, never the stores behind them.
Built to be economical: ordinary processors where they suffice, fast storage for intelligence in demand, specialised hardware only where measurements justify it, and capacity added only on evidence.
How the technology is held to account.
Facts, relationships and hypotheses kept apart
An observed fact, an inferred relationship and a hypothesis are recorded as three different things, and never presented as one another.
A source and a time on every figure
Every figure can be traced to where it came from and when it was read.
Refuse rather than guess
When a result cannot be computed, the answer says so instead of offering an estimate dressed as a result.
Intelligence, not advice
Panbiota describes, calculates and checks. The decisions are yours.
Asked and answered.
How does Panbiota work with frontier AI models?
The model asks, Panbiota answers with sourced and checkable intelligence, and the model reasons and replies. Panbiota does not build language models and is tied to none; a model may propose a question or a hypothesis, but it cannot certify a Panbiota result.
What does 'compute before the question' mean?
Panbiota is designed to ingest events continuously, find and verify relationships, keep their history and evidence, and update them as new information arrives, so that when an AI agent asks, the answer is retrieved rather than worked out from scratch. Complex new questions may still need fresh computation. For Solana this architecture is proposed, not yet built; the public instruments already work this way at small scale.
How will Panbiota's infrastructure run?
In three operational layers: ingestion, which collects events continuously; intelligence, which discovers, verifies and indexes relationships and keeps their history, updated incrementally; and delivery, secure low-latency APIs that serve specific answers. The design is proposed and awaits approval; only the delivery layer's open access point is live.
What is mathematical discovery?
Methods for finding useful relationships in data and keeping them as reusable mathematical structures that can be checked, combined and used again. At Panbiota it is in development, applied first to Solana, and a relationship is kept only if it holds beyond the data it was found on.
How does Panbiota verify its outputs?
Every figure carries its source, its licence and its time, and the daily notes are computed from recorded figures by fixed rules, so the same inputs give the same result. An exact core that checks consistency and calculates with a full trace is demonstrated in research, and is in development as a service.
Is any of Panbiota's technology live today?
Yes: the data layer of 48 live instruments, provenance on every figure, the daily notes, and the open access point for AI assistants at panbiota.ai/mcp. Discovery, verification services and secure paid access are in development, and are labelled so throughout the site.
What does the Panbiota API sell?
Answers. Panbiota's API sells answers, not the underlying assets: AI agents and software pay for specific, sourced and checkable answers, while the engine, the relationships, the history and the discovery behind them stay private to Panbiota. Paid access is in development, and terms are agreed in writing.
Does Panbiota use third-party AI models?
Panbiota's own systems do the computing. Where a Panbiota service uses a third-party language model, for example to read or write language, Panbiota says so, and what Panbiota computed is kept apart from what the language model wrote.
Panbiota provides the intelligence behind the answers.
Intelligence, not advice. The decisions are yours.