The
company
Mantikon is a decision system built at Haydamax Laboratory in Tallinn. It began in October 2026 as a written theory, a formal model and a protocol, and is now a research programme and a platform in development, selecting its first deployments.
Origin
Haydamax Laboratory is the division of Haydamax OÜ that starts companies, tests hypotheses, and builds and launches new services for the group. Mantikon is one of its hypotheses: that the decisions large organizations struggle with most are short not of data or of intelligence, but of a way to hold every interest at full strength at the same time, on a record that can be audited, behind a gate that no model can talk its way through.
The laboratory's answer is a council of specialized models rather than a single model, and a protocol rather than a prompt. The council is demonstrated on this site. The theory and the research programme are written in a paper dated 11 October 2026. The platform around the council, the data fabric that feeds it and the shield that controls what it may do, is in development.
The project is aimed at the market that Palantir, and after it IBM, Microsoft, Dataiku and C3 AI, have defined: organizations whose decisions move plants, fleets, grids, hospitals and crews. Mantikon does not claim to be the only system that sees data, deliberates and acts; it claims a narrower set of measurable properties, and it intends to prove them in comparable pilots rather than in a brochure.
Written, demonstrated, in development, selecting.
The project is reported as it is. Nothing here is a customer, a metric or a promise; the stages are what exists today, 11 October 2026.
- 005A-01
Theory and protocol
The definition of Digital Consciousness, the eight requirements, the formal model, the ten roles of the council and the fifteen phases of MDP/1, eight falsifiable hypotheses and the evaluation design.
Written - 005A-02
The council
A scripted council arguing three illustrative cases to a verdict with its dissent, running on this site. A demonstration of the protocol, not an implementation of the engine.
Demonstrated - 005A-03
Laboratory prototype
Versioned arguments, the baselines, the metrics and the evaluation harness. The first experiments of the programme.
In development - 005A-04
Enterprise MVP
A council for CRM and commercial operations in modes A1 and A2, with a data fabric, a dossier, approvals and a sandboxed write-back.
Planned - 005A-05
First deployments
A small number of organizations whose decisions are worth the council's first real cases, and whose data can be back-tested without leaking the future.
Selecting
The name
Mantikon is the Greek μαντικόν, "the prophetic": the faculty of seeing forward. Plato uses the word for the part of the soul that foresees, and for the art of the oracle whose sayings had to be weighed by interpreters before they became decisions. That is the model: the council is the oracle, the organization is the interpreter, and the decision is what they settle on together.
The eight capitals of the name are the same in Latin and in Greek, ΜΑΝΤΙΚΟΝ, which is why the K stays. A second reading is the manticore, the creature with a human face; the project keeps the first.
Palantír, the stone of Tolkien's tales, sees far. Mantikon sees forward.
Who it
is for
- 01Commercial operations first. Sales and finance organizations whose CRM and ERP already hold the facts: lead priority, churn risk, receivables, budget deviations, project execution. The system prepares the report, checks the inputs and drafts correctable proposals; it does not run the company.
- 02Manufacturing, energy and logistics as a separate line of engineering validation: sensor observation, mismatch detection, failure forecasts, maintenance scheduling, scenario loads. Proposing a maintenance schedule is not changing a protective parameter; the safety PLC, the interlocks and the proven control system stay independent.
- 03Health systems and the public sector, which must explain a decision to the people it affects, and where the record of the argument is the product.
- 04Space and other crewed operations only as a separate, multi-year safety-critical programme. A council could help a crew and a control centre with diagnosis, experiment planning and the ranking of scenarios under limited contact; access to onboard commands follows from industry norms, hazard analysis and qualification, never from a CRM pilot.
What is
not claimed
That the system experiences anything. That dialogue by itself produces intelligence. That the idea of a mind as a society of agents is new. That Mantikon outperforms Palantir or any other platform; that is a matter for comparable pilots and a measured cost of ownership, and the cost of the incumbents is "requires a commercial quote", not "more than Mantikon". That any number on this site is a result: the acceptance targets are targets, and the durations are a planning scenario. The project is honest about this because the research programme is built to produce negative results as readily as positive ones, and a negative result changes the architecture and the commercial promise.
The research programme