When Intelligence Got Cheap: What Is Expensive Now
An essay on cheap applied intelligence, the reorganisation of work, and the value of intent, judgment, relationships, and new tasks.
Read the essay →Writing
Not AI news. An examination of what happens after the demo.
I write about systems that meet real sources, organisational rules, law, history, and responsibility. The subject is not model capability alone, but the quality of the decisions built around it.
A system is verifiable when a material result can be connected to a source, data version, method, test, and human decision. This is not a certificate of infallibility; it is a property of architecture and process.
02Legal AI and Computational LawLegal AI covers systems supporting work with rules, decisions, and case material. It requires explicit source currency, interpretation boundaries, and professional accountability.
03Agentic EngineeringAgentic engineering structures roles, state, tools, and quality gates around a task. Its goal is not to maximise agent count but to control accountability and artifacts.
04Open and Applied ResearchApplied research becomes a public product when the conclusion is published with data, code, assumptions, version, and limitations.
05AI, Institutions and SocietyTechnology changes institutions through concrete procedures, criteria, and accountability structures. AI analysis therefore requires attention to organisational decisions, not only model capability.
Archive
All published material
AI, institutions and society
AI Has Hacked the Code of Human Civilization
An original analysis by Sergei Audzeichyk of questions raised by Yuval Noah Harari’s lecture about agency, institutions, language, and human judgment.
Read the original analysis →Pressing Record Is Easy. Explaining It to a Regulator Is Harder
Google Meet, Zoom, Fireflies.ai, Otter.ai, and Gong can place a single conversation on a path through several systems—turning it into a transcript, an AI summary, a CRM record, and part of an organisation’s memory. In 2025, the French data protection authority fined a call-centre company €250,000 for infringements involving data minimisation, retention, and security. An interactive publication about seeing the consequences before recording begins.
Read the publication →A country concentrating around its capital
A long time series can reveal structural change, but only with explicit definitions, sources, and a clear boundary between history and scenario.
A swarm instead of a platform: a new architecture for national legal systems
An architectural thesis: legal infrastructure can use many controlled agent workspaces when they exchange verifiable work packages through a shared contract.
Law as a chain of events
An event model separates documents, facts, actions, and legal qualification instead of treating law as a flat collection of text fragments.
Codex as a personal operating environment for office work
The value of an agent workspace comes not from one answer but from a controlled project that preserves instructions, artifacts, tests, and a history of repeatable work.
A two-role Codex workflow from specification to CI
Controlled agent work starts with a contract, role separation, and an external gate—not with the number of agents.
Open source as evidence of work
Public code can reveal process, tests, and decisions, but only scope and limitations make it interpretable as evidence.
Fundamental LLM problems in legal systems
Legal LLM failures come not only from hallucinations but also from unclear corpora, temporal scope, and accountability boundaries.
Retrieval crash test: when similar means wrong
High semantic similarity can return a document that is wrong in time, procedure, or legal subject.
Legal AI copilot architecture: retrieval, graphs, and an evidence path
Comparing vector retrieval, knowledge graphs, and orchestration leads to an architecture where each tool has a bounded role and the result preserves a path to its source.
Material published only on Medium
Links lead to the complete external text; no empty internal pages are created for them.
Society after AI
An essay on the institutional consequences of automation: responsibility, labour, and the distribution of knowledge—not model capability alone.
Read on Medium →The horse, the locomotive, and the limits of technological analogy
An examination of the familiar comparison between AI and earlier technological transitions—and what that analogy conceals.
Read on Medium →Knowledge graphs and LLMs
Why knowledge structure and claim provenance still matter when the generative model itself is strong.
Read on Medium →LLMs and XGBoost: different tools for different errors
An engineering note on choosing a model for the problem structure, data, and cost of failure—not the appeal of its interface.
Read on Medium →Eurasian logistics as a data and institutional problem
A research view of a system in which routes are shaped not only by distance, but by rules, borders, and data quality.
Read on Medium →