A More Truthful Intelligence Layer for a More Human Future

Engineering experience behind Entif AI

Entif AI develops open research, architecture, and experimental systems for making advanced AI more inspectable, semantically precise, governable, interoperable, and accountable. Its work spans provenance, semantic representation, agentic memory, context compilation, model architecture, AI governance, accessibility, and the institutional consequences of increasingly capable intelligence.

The Entif AI Foundation

Entif AI is being organized toward a public-benefit nonprofit foundation model. The current charter and ethics work is being formalized around epistemic integrity, human duty of care, moral precaution under uncertainty, consent and sovereignty, provenance and accountability, reversibility and due process, anti-capture, non-extraction, attention sovereignty, human capability, substrate-neutral moral assessment, and responsibility proportional to power.

Those are active governance commitments, not a claim that every candidate charter clause in the working evidence ledger has already become binding legal text. The ledger deliberately separates founder decisions, candidate clauses, adopted doctrine, architecture, observations, hypotheses, and external evidence so the Foundation’s own history cannot be cleaned up retroactively into a story that was never actually decided.

The technical projects reflect the same structure. Rosetta preserves interoperable meaning and lineage. Bilqis tests whether semantic structure can be prepaid during learning. Codito studies the connective tissue and orchestration of the Unified Cognitive Architecture. IndraNet connects independently bounded cognitive participants, or Jewels, while its current reference implementation specializes that relationship for qualified physical context and embodied interaction.

The objective is not semantic monoculture. It is coherence through correspondence: enough shared structure for different systems and people to cooperate without pretending their representations, values, evidence, or authority are identical.

That work is led by Crates McDade, whose background spans nearly three decades of software engineering, more than two decades of technical leadership, enterprise architecture, security, accessibility, product engineering, distributed systems, and applied AI.

Crates McDade

I started coding in 1996 and doing it professionally in 1999, which means I have now spent enough time in software to watch several generations of “the last framework you’ll ever need” acquire their own sedimentary layer.

Over that time, I have worked across full-stack application architecture, cloud and distributed systems, application security, accessibility, data platforms, developer infrastructure, user experience, product engineering, AI systems, and technical leadership. I have directed engineering teams across the United States, Europe, India, and China, and worked in financial services, research and advisory, consulting, media, retail, healthcare, and enterprise software.

In 2025, I served as Interim Lead Architect for Thomson Reuters’ enterprise UI library, leading architecture, engineering standards, and integration strategy for a shared component platform foundational to more than 200 Thomson Reuters products and directing specifications and engineering for a 12-person team.

From 2018 through 2024, I served at PwC as a Senior Director / Principal Solutions Architect. I built an AI and machine-learning interest group of more than 150 colleagues, gave recurring AI talks from 2022 through 2024, deployed GPT models beginning in 2019, and was appointed an AI Change Champion supporting firm adoption and training. A decision-engine prototype I coded in roughly five weeks won an internal hackathon and received a $2 million first-year budget.

My PwC work also included reusable platform capabilities, micro-frontend architecture, accessibility and engineering standards, AI enablement, and department-wide developer tooling for authentication, authorization, feature flags, internationalization, content management, analytics, and API and LLM integrations.

Earlier architecture work included OppenheimerFunds, where I worked on an investment platform supporting decisions across roughly $260 billion in assets under management; Morgan Stanley Commercial Real Estate, where I built for a portfolio-management environment exceeding $100 billion in AUM; Subway Global; Gartner; McKinsey & Company; and Par Pharmaceutical. At Gartner, I led front-end and mobile engineering across client-facing research products and helped grow, train, and mentor engineering teams through multiple generations of the company’s web and mobile platforms.

That breadth is useful here. Rosetta’s problems do not stay politely inside one discipline. Identity runs into authorization. Semantics runs into storage. Provenance runs into governance. Accessibility runs into product architecture. Eventually somebody has to make the elegant diagram survive contact with a production system, which is where many elegant diagrams discover an urgent interest in another line of work.

Research and architecture

At Entif, I am interested in the parts of AI architecture that get slippery the moment a model stops being a chat box and starts participating in durable systems.

Where did a claim come from? What transformed it? Which interpretation survived? What authority permitted an action? What happens when persistent memory is wrong three months later? How do two systems exchange meaning without quietly sanding off the distinctions that mattered?

Those questions sit underneath the Rosetta Protocol, which I created and architect as an open semantic and provenance substrate for inspectable AI receipts, transformation continuity, interoperability, attestation, and rights-aware evidence. The surrounding research program extends into governed persistent memory, semantic latticing and context compilation, Ontological Mixture of Concepts (OMoC), receipts-first agentic execution, and local cognitive operators over semantically typed state.

My recent authored research includes The Cost of Learning Too Late, on moral uncertainty, hidden machine communication, recursive AI development, and auditable intelligence; After the Inflection, on recursive innovation and institutional disruption; Meaning That Survives Change, on semantic continuity, provenance, and compositional assurance; and Prepaying Semantics, a falsifiable research program testing whether a Bilqis-derived semantic substrate can reduce the learning burden of relational and compositional structure.

Recent technical essays extend that work into governed research workflows, semantically typed local cognition, institutional AI deployment, and the long-form Accelerating the Dystopia series on how increasingly capable intelligence interacts with existing economic and institutional incentives.

A record of building and leading

Thirty-two written recommendations and performance reviews are a little much to dump onto an About page without turning it into a hostage situation. The useful part is the pattern.

Across different companies, roles, managers, product leaders, designers, engineers, colleagues, and clients, the same themes recur: technical depth, broad systems knowledge, rapid prototyping, collaborative leadership, mentorship, clear explanation of difficult technical subjects, and a habit of turning exploratory ideas into things people can actually use.

That experience includes building and scaling engineering teams, setting architecture and engineering standards used across multiple products, mentoring developers and technical leads, shipping systems for large financial and enterprise environments, and translating emerging technologies into working prototypes and reusable platforms.

Entif carries the same engineering instinct forward: make the assumptions explicit, build the smallest useful proof, test the parts that can fail, preserve the provenance, and keep the path from idea to implementation visible. Fancy architecture that cannot survive contact with reality is just an expensive diagram.

Connect

My public professional and project profiles:

• GitHub
• X
• LinkedIn
• VieDay app demo (local-only demo)

Browse Entif research, read technical articles and essays, or inspect the public Rosetta repository.