Skip to main content

Verifiable reasoning for AI decisions.

Ontologic is a proof-of-reasoning protocol that enables AI agents and distributed systems to cryptographically prove the justification behind their actions, not just the actions themselves.

Ontologic is a pre-Alpha proof-of-reasoning protocol anchored on Hedera Hashgraph. It formalizes three components of any reasoning act — rule definition, inference execution, and canonical meaning — into cryptographically bound structures called Morpheme Proofs. The result: agents can prove not just what they did, but why.

Ontologic

Ascension Hackathon

1st Place ($10K)

Hedera's official hackathon — external validation from the ecosystem

Apex Hackathon

Building Now

Feb 17 – Mar 16, 2026 — $250K prize pool

Hedera-Native

HCS Anchored

Verifiable timestamps and total ordering via Hedera Consensus Service

Why This Matters

As autonomous agents start making consequential choices—moving money, granting access, routing customers, triggering incidents—we're asked to trust decisions we can't inspect. Today's networks verify actions and state changes, but the reasoning that produced them is mostly opaque and hard to audit.

Logging "what happened" isn't enough. Teams need legible, replayable reasoning checkpoints: the rule that was applied, the inputs that were considered, the deterministic inference that produced the output, and a durable meaning record that can be independently checked later. Without that, compliance becomes paperwork theater and multi-agent coordination becomes guesswork.

Ontologic's mission is to make reasoning provenance a first-class primitive: bind a rule + inputs + outputs into a cryptographically verifiable proof that any verifier can recompute—without trusting the agent that generated it. In other words: verifiable justification, not vibes, using minimal deterministic structures designed for independent validation and durable attestation.

The Thesis

The Problem

Distributed systems verify behavior but not justification. LLMs, autonomous agents, and symbolic systems produce outputs through processes that are neither reproducible nor independently verifiable. Blockchains secure state transitions, but not the logical structures that generated them. There is a gap between action and meaning.

Ontologic's Response

Ontologic proposes a deterministic, minimal protocol to close this gap. The protocol formalizes reasoning into three sequential layers — the Triune Proof Model:

The Triune Proof Model consists of three sequential layers that together form a complete verifiable reasoning chain:
1

Rule Layer (Definition)

A rule is any declarative expectation in canonical form — a logical axiom, a domain constraint, a transformation. The rule is hashed to produce a ruleHash.

The Rule Layer corresponds to the Tarski semantic layer, defining what conditions must be satisfied. In Ontologic's CMY color system, this layer is represented by Magenta.
2

Inference Layer (Execution)

Given a rule and inputs, a deterministic procedure evaluates whether the rule holds. Inputs and outputs are serialized and hashed. The inference must be deterministic and externally reproducible.

The Inference Layer corresponds to the Peirce pragmatic layer, handling the actual computation and data flow. In Ontologic's CMY color system, this layer is represented by Cyan.
3

Meaning Layer (Attestation)

The results are published as a canonical meaning record via consensus-backed attestation. Hedera Hashgraph satisfies the requirements — total ordering, verifiable timestamps, immutability, low-latency publication — through the Hedera Consensus Service.

The Meaning Layer corresponds to the Floridi information layer, establishing semantic significance and public attestation. In Ontologic's CMY color system, this layer is represented by Yellow.

The final proof — the morpheme — binds all three layers into a single verifiable hash:

Prototype Example proofHash = H(ruleHash || inputsHash || outputsHash) This formula shows how a morpheme proof is computed: the proof hash equals the hash of the concatenation of the rule hash, inputs hash, and outputs hash. This creates a single verifiable identifier that binds together the rule definition, the inference execution, and the semantic meaning.

Two independent parties computing the same reasoning derive the same morpheme. Verification does not require trust in the agent that produced the reasoning.

What's Ahead

Building toward dynamic rule registries and the OTS/OCS framework. Instead of encoding rules in smart contracts, agents will resolve human-readable rule URIs via registries built on existing Hedera standards (HCS-1 for rule storage, HCS-2 for registries, HCS-13 for schema validation). Designing for community governance of rule taxonomies. This is the direction — not the current implementation.

Hologlass — Developing an agent harness that emits Ontologic proofs at decision points within AI agent workflows. Currently building for the Apex Hackathon sprint.

Request Early Access

Ontologic is in active pre-Alpha development. Request early access to be notified when developer tooling ships.