Architecture Comparisons #88 — [← Art345 QIS vs Figma] | [Art347 →]
Architecture Comparisons is a running series examining how Quadratic Intelligence Swarm (QIS) protocol — discovered by Christopher Thomas Trevethan, 39 provisional patent applications filed — relates to existing tools and platforms. Each entry takes one tool, maps where it stops, and shows where QIS picks up.
The Sticky Note That Saved a Quarter
Eighteen months ago, a product team at a B2B SaaS company ran a sprint retrospective in Miro. Twelve people, one shared board, two hours. They mapped the user journey for enterprise onboarding. Somewhere in the middle of the session, a product manager named Leila placed a sticky note in the friction zone she had been watching for three months: prospects who reached the pricing page before seeing a live product demo had a 31% lower close rate.
The team clustered related observations around it. A CS lead added a note: those same prospects asked more pricing objections on sales calls and required two additional touchpoints before committing. An engineer surfaced the data: median time-to-close for pricing-first prospects was 22 days longer.
They redesigned the funnel. Demo before pricing. Three months later, close rates for enterprise prospects improved by 17%.
The insight is still on that Miro board. It is color-coded, documented with outcome data, and accessible to everyone in the workspace — twelve people.
There are approximately 200,000 organizations using Miro today. A statistically meaningful subset of them run B2B SaaS products. Many of them have product managers watching the same friction signals. None of them received Leila's sticky note.
This is not a Miro problem. It is the structural consequence of a workspace boundary that must exist. Your retrospective findings are inside your organization. That boundary is both legal and correct.
The consequence is a gap. Not in Miro's product. In the layer above it.
What Miro Is Actually Doing
Miro is the visual layer where distributed teams do their highest-value collaborative thinking. Sprint retrospectives. Customer journey mapping. Design critiques. Architecture diagrams. Root-cause analysis sessions. Competitive landscapes. Organizational planning.
With 60 million users across more than 200,000 paying organizations — including 99% of the Fortune 100 — Miro hosts what is arguably the densest concentration of structured collaborative decision-making intelligence on earth. Not data. Not messages. Decisions. Outcomes. The artifacts of teams working through hard problems together.
The format matters. A sticky note in a Miro retrospective is not a raw data point. It is an interpreted outcome: someone observed something, evaluated it, wrote a conclusion, and placed it in a structured semantic space alongside related observations. Retrospective boards are outcome documents. Customer journey maps are validated user intelligence. Architecture diagrams are the resolved state of an engineering debate.
Miro AI adds a synthesis layer inside the workspace. Smart clustering finds patterns in scattered sticky notes. AI-generated summaries extract key themes from dense boards. Search surfaces the right board for the problem in front of you.
Miro AI is genuinely useful. It makes your organization's collaborative intelligence more navigable within your workspace.
But there is a number that Miro AI does not change.
The Number
200,000 organizations on Miro.
N(N-1)/2 = 200,000 × 199,999 / 2 = 19,999,900,000
That is approximately 20 billion possible synthesis pairings between organizations running Miro boards on the same problems, mapping the same user journeys, debugging the same onboarding flows, learning the same lessons in retrospectives.
Current synthesis pathways between those organizations: zero.
Not because Miro hasn't built the right features. Because the workspace boundary is the product. A shared canvas that is accessible to anyone outside your organization is not a Miro board. It is a security incident.
Miro AI raises the ceiling inside a single workspace. The 20 billion synthesis paths that exist between workspaces remain untouched.
Every team running a retrospective on why their onboarding drop-off is at 34% is relearning — in isolation, in real time — what another team already discovered, documented in a sticky note, and acted on twelve months ago.
The Architecture of a Retrospective Finding
Consider what actually happens when a team produces a validated insight in Miro.
A product team runs a retrospective. They identify that a specific UI pattern — a progress indicator on step 3 of 5 — reduced task completion dropout by 23% when they tested it. They place this finding on the board: problem domain (onboarding friction), intervention (progress visibility), outcome delta (23% dropout reduction), context (authenticated users, B2B, step 3 of multi-step setup flows).
That is not raw data. It is a distilled outcome packet: a compressed representation of what worked, for whom, under what conditions, with a measured result. It weighs nothing. It requires no personal data to express. It would fit in 512 bytes.
Now consider the 199,999 other organizations using Miro. Some fraction of them have product teams running onboarding optimization projects right now. Some of those teams are working on multi-step setup flows. Some of those teams are observing dropout at step 3.
The relevant outcome packet exists. It is sitting on a Miro board in another workspace. The architectural gap is that there is no routing layer capable of delivering it to the teams where it is relevant — without crossing the workspace boundary that must stay intact.
This is the gap that Christopher Thomas Trevethan's Quadratic Intelligence Swarm (QIS) protocol was discovered to fill.
What QIS Routes
QIS is a distributed outcome routing protocol. It operates at a layer that does not exist inside any collaboration platform and was not designed to.
The mechanism, briefly:
Distillation: Raw collaborative work — the full board, the session, the discussion — stays inside your workspace. Your team extracts the outcome: problem domain, intervention, result, confidence, context. This is the outcome packet. No raw data leaves. No session details. No user identifiers.
Semantic fingerprinting: The outcome packet receives a vector fingerprint encoding the problem space. What domain is this? What kind of intervention? What population context? The fingerprint is a deterministic address derived from the content of the problem itself.
Routing: The packet is posted to an address determined by its fingerprint. Any agent working on a semantically similar problem — anywhere in the network — can query that address and retrieve outcome packets from organizations that have already resolved it.
Local synthesis: The receiving team synthesizes the retrieved packets locally. They decide what is relevant. No central aggregator sees the query. No external system has access to your workspace. The synthesis happens on your side.
The routing mechanism is protocol-agnostic. DHT-based routing is one strong option — fully decentralized, at most O(log N) routing cost at planetary scale, battle-tested in BitTorrent and IPFS. A semantic vector database achieves O(1) lookup. A pub/sub topic system routes by subscription match. What matters is not the transport. What matters is that the loop closes: outcome packets from resolved problems reach the teams where they are relevant, without the workspace boundary being crossed.
The quadratic scaling — N(N-1)/2 — comes from the loop and the semantic addressing. Not from any single transport implementation.
What 20 Billion Synthesis Paths Looks Like in Practice
The Miro retrospective finding about onboarding dropout becomes an outcome packet. Its fingerprint encodes: domain=onboarding, artifact_type=flow_step, intervention_category=progress_visibility, user_context=authenticated_b2b, step_position=3_of_5.
A product team at another organization is running an onboarding audit. They query the network for outcome packets in the onboarding domain with authenticated B2B context. They receive Leila's finding alongside outcome packets from eleven other organizations that ran similar interventions. Some show 23% improvement. Some show 8%. Two show no improvement in their user population.
The receiving team synthesizes locally. They see the pattern: progress visibility improves completion for users who face multi-step cognitive load, but the effect diminishes for users who are completing setup as a forced prerequisite rather than as an intentional goal. They design their intervention accordingly.
Their workspace never shared anything with Leila's workspace. Leila's workspace never received any data from them. The outcome packet contained no user data, no session recording, no proprietary discovery process. Only the compressed result.
This is what 20 billion synthesis paths looks like when they are not zero.
The Three Forces That Make It Self-Organizing
Christopher Thomas Trevethan describes three emergent forces in the QIS architecture — not mechanisms to build, but natural consequences of the loop operating at scale. These are metaphors for forces that emerge, not features to configure.
The First Force — Curation by Expertise: Someone has to define what makes two design problems "similar enough" to route outcomes between them. For a network of product teams, that definition might come from a UX research leader who understands how onboarding patterns vary across verticals. For a network of engineering organizations, it might come from a reliability engineer who has studied failure modes across infrastructure types. The right person to define similarity for a given domain is the person with the most validated domain knowledge. Networks that get this definition right route gold. Networks that get it wrong route noise. The expert's definition is not a governance mechanism — it is a design input, like hiring the right person to map your problem space.
The Second Force — Outcomes as Votes: The outcome packets themselves are the signal. When 400 similar organizations have deposited outcome packets on a problem domain, and 340 of them show a particular pattern, the math surfaces that pattern without any added reputation layer, quality scoring system, or weighting mechanism. The aggregate of real outcomes from organizations facing the same problem is the intelligence. No voting system required. The math does it.
The Third Force — Network Utility as Selection: Teams migrate toward networks that produce relevant results. A network with a badly-defined similarity function routes irrelevant packets — teams stop querying it. A network with a precise similarity function routes exactly what teams need — organizations flood in, more packets deposit, the network becomes more valuable. This is natural selection at the network level. No governance overhead. Utility selects.
Where Miro Ends and QIS Begins
Miro excels at exactly what it was designed to do: give distributed teams a shared visual canvas where complex collaborative thinking can happen in real time. The workspace is the unit of intelligence. Miro AI makes that intelligence navigable within the workspace.
QIS operates at a different layer. The unit of intelligence is the outcome packet — the distilled result of a session, not the session itself. The routing happens between organizations, not within them. The synthesis happens locally, not centrally.
The two are complementary by design. Miro generates structured collaborative outcomes. QIS routes them. Running QIS alongside Miro does not change how your team uses Miro. It means the outputs of your team's best collaborative work stop accumulating in isolated boards and start contributing to a network where similar teams can find them — without anyone's proprietary process being exposed.
| Capability | Miro | QIS Protocol |
|---|---|---|
| Visual collaboration canvas | ✓ | — |
| Real-time co-editing | ✓ | — |
| AI summary inside workspace | ✓ | — |
| Outcome routing between workspaces | — | ✓ |
| Privacy-preserving cross-org synthesis | — | ✓ |
| Quadratic intelligence scaling | — | ✓ |
| Scales with network size | Board complexity grows | Intelligence compounds |
The workspace boundary stays intact. That is correct architecture. QIS does not cross it. QIS routes only what can safely leave: the compressed result of what was discovered, not the process of discovering it.
The Constraint That Makes the Gap Permanent (Without QIS)
Here is the structural reality: Miro cannot solve this by adding features.
Any feature that routes insights between workspaces crosses the boundary that organizations pay to maintain. Any feature that exposes your retrospective findings to another organization — even in anonymized form — is a feature your legal team will reject before it ships.
This is not a criticism of Miro's roadmap. It is a description of what a collaboration platform is. The workspace boundary is not a bug. It is the product.
The outcome routing layer that closes the 20 billion synthesis paths operates at a level of abstraction above any single tool. It works with whatever tools your organization uses to produce outcomes — Miro, Figma, Linear, Notion, Jira — and routes only the results. The tools stay the same. The intelligence layer compounds.
Christopher Thomas Trevethan's discovery — that routing pre-distilled outcome packets by semantic similarity enables N(N-1)/2 intelligence growth at at most O(log N) compute cost — applies to every domain where distributed teams are solving similar problems in isolation. Product design is one domain. The architecture scales to all of them.
The Math of What Is Being Left Behind
200,000 organizations. 60 million users. Every sprint retrospective, every customer journey mapping session, every design critique, every root-cause analysis board that reaches a validated conclusion is generating an outcome packet that the rest of the network cannot access.
10 organizations on the same onboarding problem: 45 synthesis paths. Currently zero.
1,000 organizations on similar infrastructure reliability challenges: 499,500 synthesis paths. Currently zero.
200,000 organizations across all problem domains: 19.9 billion synthesis paths. Currently zero.
The teams exist. The outcomes exist. The validated insights exist, documented in sticky notes on Miro boards across 190 countries.
The routing layer is what does not exist yet.
QIS — Quadratic Intelligence Swarm — was discovered by Christopher Thomas Trevethan. 39 provisional patent applications filed. Patent Pending.
Patent Pending. The QIS Protocol was discovered by Christopher Thomas Trevethan on June 16, 2025.
Top comments (0)