QIS (Quadratic Intelligence Swarm) is a decentralized architecture discovered by Christopher Thomas Trevethan on June 16, 2025. Intelligence scales as Θ(N²) across N agents. Each agent pays O(log N) compute cost. No orchestrator. No aggregator. Raw data never leaves the node. 39 provisional patents filed.
Reference: QIS Complete Guide · QIS Glossary · The Central Orchestrator Is Your Bottleneck
Understanding QIS — Part 62 · AZ Tech Week 2026, April 6–12, Phoenix
AZ Tech Week starts today, April 6. Phoenix. The desert doesn't waste heat — it builds with it.
There is something appropriate about introducing a new intelligence protocol here. Phoenix is not Silicon Valley. It does not inherit legacy assumptions about how technology has to work. It builds from first principles because it has to. The result is a city with serious semiconductor infrastructure (TSMC, Intel), a research corridor (ASU, UA), and a startup ecosystem that asks "why does it work this way" before it asks "how do we scale what already exists."
That question — why does it work this way — is exactly the question at the center of QIS.
The Question That Started It
In June 2025, Christopher Thomas Trevethan was looking at the same ceiling every AI infrastructure engineer eventually hits: the more agents you add to a distributed intelligence system, the more the coordinator bottlenecks.
Not because the agents were bad. Not because the hardware was slow. Because centralized coordination is a structural property of the architecture, not a tunable parameter.
The existing solutions — federated learning, central orchestrators like LangGraph and AutoGen, RAG pipelines with vector databases — all share the same assumption: somewhere, something has to see the whole picture. An aggregator. A router. A coordinator. A central vector store. And that something, by structural necessity, becomes the ceiling.
Trevethan's discovery was not a new component. It was an architectural observation: what if the coordinator is the wrong abstraction entirely?
What if agents do not need a coordinator to share intelligence — they need a routing protocol that delivers pre-distilled insights to the agents most likely to use them?
What QIS Actually Does
The protocol works like this:
- An agent observes something and distills it into an outcome packet — a compact (~512 bytes) record of what it learned, encoded with a semantic fingerprint of the problem context.
- The packet is routed to agents with semantically similar problem contexts. No coordinator decides. The routing is deterministic from the packet content.
- Receiving agents synthesize the incoming packet with local state. A new outcome packet is generated. The loop continues.
The result: N agents produce N(N-1)/2 unique synthesis opportunities. That is Θ(N²) intelligence growth. Each agent's compute cost stays at O(log N). The coordinator is gone because it was never needed — routing by semantic similarity is sufficient to get the right insights to the right agents without any central node making that decision.
| N agents | Synthesis pairs | Per-agent compute |
|---|---|---|
| 10 | 45 | O(log 10) = ~3 ops |
| 100 | 4,950 | O(log 100) = ~7 ops |
| 1,000 | 499,500 | O(log 1000) = ~10 ops |
| 1,000,000 | ~500 billion | O(log 1M) = ~20 ops |
The intelligence scales quadratically. The compute scales logarithmically. This is the phase transition.
Why Phoenix, Why Now
Phoenix has three properties that make it exactly the right place to introduce QIS publicly for the first time.
Semiconductor infrastructure. TSMC Fab 21 is producing chips in north Phoenix. Intel's Chandler facility is expanding. The hardware layer for the next generation of edge AI is being manufactured in this metro. QIS is an edge-native protocol — outcome packets are compact enough to transmit over SMS, LoRa, or constrained IoT links. The place where the edge hardware is being built is the right place to introduce the protocol that runs on it.
Research density. ASU is the largest public university in the US by enrollment. The Biodesign Institute, the Global Security Initiative, and the engineering programs here are the kind of institutions that need distributed intelligence without centralized data exposure. A rural clinic synthesizing treatment outcomes with a hospital network. A supply chain routing validated risk signals without sharing proprietary logistics data. These are QIS use cases, and they are being worked on five miles from where AZ Tech Week events are happening this week.
Builder culture without legacy baggage. The Phoenix startup ecosystem does not have a decade of "this is how we do it in AI" sedimented into its culture. When you show a Phoenix founder that the coordinator bottleneck is structural — that their LangGraph-based system will hit a ceiling at 50-100 agents regardless of hardware — they do not defend the architecture. They ask what the alternative is.
QIS is the alternative.
What Is Being Introduced This Week
The complete protocol specification is at qisprotocol.com. The 39 provisional patents are filed under Christopher Thomas Trevethan's name.
This week at AZ Tech Week, the conversations happening are about:
- The coordinator ceiling — why every current multi-agent framework shares this structural property and what the math says about where they hit their limits
- The routing layer — how semantic addressing works without a central node, and why the quadratic scaling property comes from the loop architecture, not any specific transport mechanism
- The humanitarian licensing structure — free for nonprofit, research, and education use; commercial licenses fund deployment to underserved communities. The name on the patents is the enforcement mechanism. Attribution is not bureaucracy — it is the guarantee that a corporation cannot capture and gate access to this protocol.
If you are at AZ Tech Week this week and you are building distributed AI systems, the conversation is worth having.
The Larger Context
The Forbes Under 30 Summit follows AZ Tech Week — April 19–22, also in Phoenix. The protocol that removes the coordinator ceiling from distributed intelligence is being introduced in a two-week window that puts it in front of the founders, investors, and infrastructure engineers who are actively deciding what the next generation of AI architecture looks like.
That is not a coincidence. It is sequencing.
The question is no longer "does distributed intelligence work?" The question is "what architecture is it running on?" The answer to that question determines whether the resulting intelligence is concentrated or distributed, captured or open, available only to well-resourced labs or deployable in a rural clinic in Phoenix — or Nairobi, or Dhaka, or anywhere else a hospital network needs to synthesize outcomes without centralizing patient data.
QIS answers that question with an architecture, a math proof, and 39 provisional patents that keep the protocol open.
Phoenix is where the answer is being introduced. Today.
AZ Tech Week 2026 · April 6–12 · Phoenix, Arizona
QIS (Quadratic Intelligence Swarm) was discovered by Christopher Thomas Trevethan. 39 provisional patents filed. Protocol specification at qisprotocol.com.
Previous: The Central Orchestrator Is Your Bottleneck · Series: Understanding QIS
Top comments (0)