Chris Kincade's quarter-century in media and technology spans work with major publishers including McGraw-Hill, the design of platforms for social bookmarking and news personalisation, and the development of frameworks underlying Cognitive Design. At Starling Memory Works, these threads converge into something unexpected: an "Org Brain" built from two and a half decades of business blueprints.

The platform emerged from what began as operating manuals for AI systems, evolved into an AI-agnostic knowledge repository, and finally became a permanent, governed memory that any AI can draw from. The result: artificial intelligence that actually comprehends the business, and the business retains ownership of what it knows.

Kincade introduced the Hierarchy of Organisational Values at a January 2023 summit alongside Ari Meisel, spending the following two years refining the concept with clients before tailoring it for artificial intelligence. He personally wrote Starling's operating system in English rather than code before assembling an engineering team to transform it into a working platform. The team includes CTO Jeremy Decker, formerly responsible for backend infrastructure at Mastercard; divisional leaders Zuhayr Tariq, Ahmed Khalid and Andrey Gubanov; and Harol Garcia, who joined in June to establish the company's own repository.

The Problem and Why Existing Solutions Don't Work

Every AI system available today operates from a shared premise: the intelligence resides within the machine, accumulating inside a platform beyond your control.

According to Kincade, the solution is organisational rather than technical. "The fix is organisational, not technical. We have to decouple the AI system from our memory, which includes daily work (i.e., active memory). Whatever the model needs to provide contextualised assistance has to live in a governed, plain-text repository the organisation owns. Any 'permissioned' AI can read and write to it, but the repository has to be so well-wired that retention is not even necessary."

This necessity for a fundamentally different approach stems from how the industry currently connects AI to knowledge. Most systems rely on derived copies—embeddings, vector databases, retrieval pipelines—which are essentially shadows of the truth that begin degrading the moment the original information changes. Starling eliminates this category entirely. The library itself functions as the cache. Each session assembles context fresh from the canonical source. "Read, reason and forget – the next session reads the same source again. Nothing goes stale, because there is no copy to drift against. There is only the source of truth."

Restraint Through Architecture, Not Petition

Kincade has characterised most AI leadership as suffering from an "Oppenheimer complex." This summer, a thousand AI workers, including well-known figures, submitted a petition to Washington requesting regulation of their own field. "When the people building a technology ask the government to restrain them, they're admitting they haven't built restraint into the technology itself."

His alternative is to keep machines accountable through design and architecture—which in practice means humans control every critical decision point. Nothing the AI produces becomes organisational truth until a person approves it. Every response indicates its source and who verified it. A written code of conduct defines what the AI can and cannot do.

"The AI can do the drafting and the finding; only people do the deciding. That's what putting humans at the centre actually entails. It's not a policy statement, but switches only people can flip."

Universal Cognitive Architecture and Open Standards

Starling is releasing Universal Cognitive Architecture as a free standard, a choice rooted in principle. Governance and the capacity to own organisational memory should not be locked behind proprietary walls. The transformative standards throughout history remained open: the Dewey Decimal System never licensed libraries; HTML did not paywall the web; Git did not charge for commits. UCA assigns every piece of organisational knowledge a permanent, human-readable address, becoming genuine infrastructure only when anyone can build on it without seeking permission.

The standard includes instructions for integration into Claude projects. Kincade envisions it enabling an entirely new profession: Cognitive Designers capable of structuring organisational memory for human governance. Early builders already deployed their own systems using the published framework before any commercial launch—precisely the intended outcome. "If UCA only ever powered Starling, we'd have a killer app. Our ambition is a retrieval and assembly standard people adopt because it's fast and effective, not just sovereign."

The deeper vision challenges the current trajectory of artificial intelligence. "Honestly, if you follow the logic all the way through, we're pursuing an alternative to AI having instant access to all human knowledge. That's what the massive data centres are needed for. The DLM solves the big compute question on an organisational basis. With it, companies find their knowledge the same way whether they have 28 documents, or 28,000 – by address, not search."

Conventional data centres become necessary when synthesising across knowledge bases built on SaaS workflows, managing dozens of document types, hundreds of APIs and thousands of applications. UCA introduces an AI-native workflow where knowledge resides in lossless markdown. Conversion happens only when information leaves the organisation.

Knowledge Inside AI Platforms as Operational Risk

Storing knowledge inside proprietary AI platforms creates genuine operational hazards. When AI trains on your proprietary information, your distinctive values and procedures become part of the new standard baseline. Service Level Agreements promising "no training and storage" offer limited protection when the AI lacks independent memory.

Two incidents this summer illustrated these dangers. In June, a government directive removed frontier models from worldwide access overnight. During the 18 days required to restore access, any working knowledge stored inside the platform became unreachable. Separately, a security team responding to an active crisis requested assistance from a commercial model and received refusal due to the model's built-in safety restrictions.

The first represents a memory risk: knowledge you do not control can be withheld from you. The second represents a cognition risk: leased intelligence can refuse you. A Domain Language Model withstands both scenarios. Kincade defines a DLM as stateless, session-based, and redundant AI systems reading and writing to a sovereign repository. "If any one AI system is unavailable, the system can switch to another."

Source: TechRound