The Right Context, Not All of It
Written deliberately as a lesson rather than a product story, taught through The Curator v3.80.0.
Everything I have published: essays and manuals from From Lab to Life, research papers, and the archive. Each work tells you where you can read it.

Start here
Written deliberately as a lesson rather than a product story, taught through The Curator v3.80.0.
Three kinds of context, three write rules, and a project that outlives the session, the harness, the model and the machine
Building a modern website in the age of AI agents: a manual, told through the new website of The Curator
The Curator ships as a native Mac app (v3.45.0). Covers the three layers (personal wiki, Shared Brain, agent working state), the menu-bar widget, bring-your-own-key across three providers at about €5 a month, and plain-markdown portability.
The confession piece about two years of hand-written handoff files. The flaw: the most degraded model writes the most important document.
I turned fifty-plus articles into a corpus you can question, download, and hand to your own AI.
New essays from From Lab to Life arrive on Substack before they reach this library.
Can a frontier orchestrator on a subscription delegate to local open-weight models?
Seven months, two generations, one production rebuild — and five things I learned that I did not expect to learn.
The behind-the-scenes rebuild of Lumina from Gen-1 to an enterprise-grade Gen-2.
A summer manual for the curious non-developer who keeps seeing people build apps out of thin air and wondering: could I actually do that?
Running Lumina with real customers six months after launch. A Ljubljana massage studio hit a forgotten 2,000-character personality limit mid-conversation; the systemic fix raised the limit fourfold and added live counters.
Putting the Organisation of Tomorrow to Work; Use Cases, Vendor Independence, and the Brain Behind the Framework
A follow-up seven weeks after the Shared Brain launch. The author's own wiki has 3,305 nodes and 15,095 edges at v3.0.6.
Two Anthropic announcements 24 hours apart. Sonnet 5 (30 June) becomes the new default, and its real story is follow-through on multi-step tasks.
We are learning to build and ship real digital products using AI coding agents. The gap between idea and shipped product is no longer technical, it is contextual.
How Claude Code plus Playwright via MCP became an autonomous testing loop in early 2025, before "loop engineering" had a name.
Every founder I know has the same problem with email support. In the early days, you handle it yourself. You are fast, you are personal, and you actually know the product. Then you get busier. Response times slip. Users start to notice. You think about hiring someone, but the volume does not justify it yet. So you buy a helpdesk tool, set up some canned responses, and tell yourself you will get to the backlog this weekend. You never get to the backlog.
Compares Kimi K2.7 and Claude Fable 5 using real production data (24 landing pages, 12 prompts).
Release notes for The Curator v3.0.1 beta.20 to beta.23: Shared Brain synchronisation improvements, a new test-automation infrastructure, and stability fixes.
Inside the machinery that powers modern AI development, and why understanding it matters even if you never write a line of code
In the first article, I named the phases: Context Engineering, The Build Phase, Production and Deployment. That naming had an unintentional effect. By calling Phase One “Context Engineering,” I implied that context engineering was something you did once — at the beginning — and then left behind.
ORGANISATION OF TOMORROW (ØØT): An Open-Source Framework and Playbook for Partner-Run, AI-Augmented Organisations
Introduces The Curator v3.0.0-beta Shared Brain: private, attributed contributions merged into a collective wiki for teams and cohorts.
A survey of how the industry is attacking agent memory (vendor memory products, vector stores, managed memory services, file-based memory), why context windows alone fail at agent scale ("context rot"), and why a compiled Markdown…
State of AI, May 2026. Frontier models have reached capability thresholds (1M context, open-source parity).
How My Curator MCP Turns Your Second Brain into a Frontier AI Research Partner
The philosophical case for the second brain: knowledge immortality, externalised expertise, and curation (compounding) versus retrieval (stateless).
Introduces The Curator through Karpathy's "LLM as wiki" insight: a local, private, Markdown-based knowledge graph that Obsidian can read.
A practitioner's comparison of three coding-agent philosophies.
The canonical home of the Solo era / Symphony era framing.
Two years ago, I started an experiment. I wanted to know whether I could build production-grade software products without writing most of the code myself. Not because I was lazy. Because I had thirty years of product ideas and not nearly enough development hours to explore them.
Five days testing Augment Code's Intent orchestration tool on the production Lumina platform (building Lumina Pro).
It all started with a conversation between me and my friend Anton Dobrina — legal expert, fellow AI enthusiast, and someone who has spent years studying and working in legal practice in Slovenia, co-writing Slovenian legislation and is an expert in the Slovenian retirement code.
How Augment Code’s Intent Just Changed Everything — Again
A research-based overview of OpenClaw (formerly Moltbot / Clawdbot), the open-source, always-on personal AI assistant created by Peter Steinberger.
An 18-month build of a legal-document AI for Slovenian building legislation (Moj AI).
An AI multi-agent marketing workflow in which different agents handle different content types.
The Lumina Widget build story: a 30-day sprint from concept to production on a budget of about $100 in AI credits, using Augment Code with Claude Haiku 4.5.
How the first version of Lumina was prototyped in Google AI Studio and taken to production on Firebase in 48 hours.
A Two-Year Journey From Fiction to Reality (and the Creativity Blockers Still Holding Us Back)
The foundational definitions essay. Sets out the taxonomy of chatbots, assistants, agents and autonomous workers, with a four-tier autonomy hierarchy and a five-level maturity model for agentic AI.
The MountVacation MCP server build story, and one of the first open-source MCP servers for tourism.
A personal narrative of the shift from "learning to code" to "directing intelligence", tracing the tools from early agents to orchestration.
In my work, I regularly encounter businesses paralyzed by AI’s data requirements. A law firm wants to analyze contracts using AI but cannot risk client confidentiality. A manufacturer needs AI-powered documentation analysis but won’t send proprietary processes to external servers.
Mastering the discipline that transforms how we communicate with artificial intelligence.
My reflections on navigating humanity’s greatest transformation.
A practical toolkit for context engineering at every skill level. Covers model-selection frameworks, prompt patterns, and how to match a tool's capability to the complexity of the use case.
A complete non-technical manual on LLMs. Covers how they work and what they are not, the training process, in-context learning, and the "stochastic parrot" debate.
What happens when you give an entry-level AI model the right tools and see if it can build complex software? That’s exactly what I tested in my latest research, and the results reveal something fascinating about the path toward Artificial General Intelligence (AGI).
A short leader-facing digest of the peer-reviewed COTRUGLI research on generative AI's resource footprint (B15).
The Claude Desktop coding-agent experiment. Documents the real productivity gains alongside the quality constraints: what worked in conversational programming, where human oversight was still required, and why context management was the deciding factor.
Digital communication has become most relevant in the past decade as digital data has become extremely valuable. Communication in the form of emailing, messaging and data file transfers form the foundation behind most business transactions.
The article explores the state of legacy communication platforms that we use and depend on every day while proposing a new alternative. It offers insight into the work of Dr. Režun and his team, spanning the past five years, as they innovate a novel approach enabling immutable, non-custodial, permissionless, and decentralized communication and data ownership retention.
Blockchain technology has the potential to change the narrative and the core concept of traditional digital human communication and data management (i.e. email, messaging, data file transfers, data file management, etc.); from custodial to non-custodial; from centralized to decentralized; from permissioned to permissionless, and; from free but mined to payable but yours. Enters 4thTech with L1/L2 secured “on-chain” communication infrastructure framework powered by a dedicated SDK (i.e. software development kit).
Marks the completion of the first two seasons of the Blockchain Recorded podcast (co-founded 2021).