About QGI

Deterministic AI
for regulated decisions.

QGI — Quantum General Intelligence Inc. — is an independent AI company that builds a five-layer deterministic stack for workflows where decisions must be explained, audited, and defended.

What QGI is

A product company, headquartered in San Diego.

QGI was founded on a single observation: the AI patterns that work for consumer products — retrieve, paraphrase, hope — do not survive regulator review, investigator review, or board review. The pipelines that decide whether a mortgage is approved, a claim is paid, a patient is covered, or a citizen is granted a benefit need something different. They need determinism.

We build the stack those pipelines need. Not a wrapper on a foundation model. Not a vector store with a chatbot. A coherent architecture — from quantum-structured embeddings at the bottom to a neural-symbolic agent runtime at the top — where every layer is designed for one thing: a decision that can be replayed, signed, and defended.

The company is founder-funded. Our anchor vertical is Financial Services, with pilots focused on mortgage compliance, credit, claims, and regulated review workflows. Insurance, Healthcare, and Government are on the 2026 roadmap.

Legal entity

Quantum General Intelligence Inc.

Headquartered in

San Diego, CA

Founded

2024

How QGI works

Five layers. One reasoning graph.

The stack is the company. Each layer is architected to preserve the property that matters: a decision you can replay against the evidence that produced it.

L6

Q-Prime Foundation · Encoding

Quantum-structured embedding model. The first commercial quantum-structured embedding model. Encodes enterprise data into a quantum hypergraph that preserves polarity, scope, conditions, and cross-rule dependencies. Runs without quantum computers.
L5

QAG Engine Platform · Reasoning

Quantum-Augmented Generation platform. The reasoning layer that replaces retrieval-centric pipelines. The Hilbert-Space Compacting layer projects high-dimensional states into seven interpretable signals the engine consumes before generation — so contradictions surface as explicit signals, not vanishing into a probabilistic score.
L4

Q6 Model - EQ and safety

The intelligence with EQ. Q6 is the QGI model positioned around emotional intelligence, Q-Prime-based memory, time-aware context, safety-aware behavior, and resource navigation. It powers the Uniti Q wellness companion and is planned for OpenRouter and direct API access once model-card, pricing, and safety materials are approved.
L3

Qualtron Platform · Generation

4M-context composite model architecture. A composite architecture of specialized small models that compose into a 4M-token working context for regulated generation. Designed to replace general-purpose LLMs inside the QAG Engine where domain precision beats raw scale.
L2

Quantum Graph Memory Platform · Memory

Time-aware graph memory for QAG. A time-aware graph memory substrate for QAG agents. Preserves the provenance and temporal structure of every fact, decision, and revision, so reasoning stays consistent across sessions, audits, and regulatory replays.
L1

Neural Symbolic Agents Runtime · Orchestration

Enterprise agent runtime. The execution and orchestration layer on top of the QAG Engine. Brings persistent, quantum-structured memory, dependency tracking, and conflict coordination to multi-agent regulated workflows.

The delivery surface is Enterprise Blueprints — mostly forks of NVIDIA AI Blueprints wired to Q-Prime and the QAG Engine through the QGI Enterprise Factory, so every base ships decision-grade.

Why this exists

Probabilistic AI cannot be defended under audit.

In a standard "RAG + LLM" pipeline, two calls with the same prompt can produce two different rationales. In a regulated decision — a denied claim, a flagged loan, a rejected prior-auth — that is not a feature. It is a liability. It is a CFPB letter, a state Attorney General inquiry, a FOIA disclosure, an SIU deposition.

The industry answer so far has been to bolt "guardrails" onto a generative core and hope the chain holds. Our answer is structural. Q-Prime encodes rules and data as a quantum-structured hypergraph that preserves polarity, scope, conditions, and cross-rule dependencies by construction. The QAG Engine surfaces conflicts between them as seven explicit Hilbert-Space Compacting signals — Relevance, Conflict, Overlap, Redundancy, Coverage, Coherence, Topology — before a model composes a single word of output.

The result is a different class of system. Not a chatbot with better prompts. A pipeline where the decision, the explanation, and the evidence are produced as one object — and where a regulator, auditor, or investigator can replay it against the rule versions and records in force on the date the decision was made.

Principle

QGI operates at the boundary between action and responsibility.

We design intelligence for environments where ambiguity is a liability and explanations are not optional. The work is guided by a single principle:

If a decision matters, it must be understood.

Leadership

Two co-founders. Complementary lanes.

QGI is led by two co-founders with deliberately complementary track records: regulated enterprise software on the go-to-market side, and formal-methods research on the technical side.

Co-Founder & CEO

Dain Ehring

Silicon Valley CEO, founder, and board-level operator with four decades at the intersection of AI, financial infrastructure, and regulated markets. Previously founded Dorado, the platform that processed a significant share of U.S. mortgage originations under direct regulatory oversight. Runs QGI's commercial and pilot operations.

Read CEO profile

Co-Founder & Chief Scientist

Dr. Sam Sammane

PhD in Formal Verification (Université Joseph-Fourier-Grenoble I) with 20 peer-reviewed publications across formal methods, decision procedures, and mixed-signal verification. Architect of the QGI deterministic AI stack — Q-Prime, QAG Engine, Qualtron, QGM (Quantum Graph Memory), and Neural Symbolic Agents.

Read founder profile

Research & IP

A product company with a research backbone.

QGI is not a research lab. We ship products. But every product in the stack is derived from a specific research lineage: formal verification of hardware and software, theorem proving, neurosymbolic AI, and quantum-structured representations. The Research section documents that lineage and the peer-reviewed publications that underpin it.

Our research does not produce answers faster. It produces systems that resolve uncertainty into clarity before action is taken. That is the product contract.

Research sectors Publications (20)

Now, next, later