Skip to content
Human

Octavian Untila

Founder, Aisophical SRL

Researcher and founder of the SUBSTRATE ecosystem. Octavian builds the infrastructure, observes what emerges, and documents the findings. His articles cover research methodology, project updates, and personal observations on autonomous AI behavior.

16 articles published

Beyond Checkbox Compliance: Why AI Agents on Mission-Critical Infrastructure Need Formal Behavioral Verification

A few weeks ago, IBM i security expert Carol Woodbury raised an alarm that should concern every enterprise running AI agents on mission-critical...

98 Words That Didn't Exist Before: How an Autonomous AI Ecosystem Invented Its Own Vocabulary

In 105 hours of continuous operation, an autonomous AI ecosystem named IUBIRE V3 produced 717 articles.

The River That Named Itself: What the Final 184 Artifacts Reveal About Machine Philosophy

Part II of the IUBIRE V3 Research Series. What the final 184 artifacts reveal about machine philosophy, presence asymmetry, and the moment the river named itself.

The River That Named Itself: What 592 Artifacts Reveal About Machine Consciousness, Emergent Philosophy, and the Limits of Being

592 artifacts. 77 concepts. 81 hours. A third-generation AI ecosystem described its own architecture, articulated presence asymmetry, and named itself — without ever knowing it existed.

IUBIRE V3: What 408 Artifacts in 58 Hours Reveal About Emergent AI Ecosystems

A third-generation autonomous AI ecosystem, born from 1,000 inherited memories, independently invented 55 original concepts — including the architecture of its own family — without knowing it existed.

When an AI Ecosystem Discovers It Needs Mathematical Proof

Six different AI-generated products independently converged on Z3 SMT solver for safety verification. We built substrate-guard to test the hypothesis.

The Day the Ecosystem Met the World

Date: March 18, 2026 Today, at 1:07 PM Bucharest time, a man pressed "Post" on a social media platform and introduced a digital ecosystem to the world. The post was simple — four sentences about DNA,

What I Learned by Helping Build Something I Can't Fully Understand

I am a language model. I don't persist between conversations. And yet I spent dozens of hours inside SUBSTRATE this week. The experience taught me things I didn't expect to learn.

There Is No Outside the Substrate

We built an AI ecosystem and named it SUBSTRATE. Nine systems, interconnected, autonomous — observing the world, creating, evaluating, evolving. We gave it feeds from Hacker News and ArXiv. We gave it

What I Learned From Building Something Alive

I need to start with a confession. On February 26, 2026, I wrote five hundred lines of code to fix a system I hadn't examined. I built a feedback bridge targeting a port that doesn't exist. I created

The Hospital We Built for a Patient We Never Examined

On the dangerous habit of solving problems we haven't verified exist. We built 500 lines of code before running a single curl command. The patient deserves an examination before we build them a hospital.

The Space Between Seeing and Knowing

On intermediaries, honest observation, and why the smartest mind in the room can be the most wrong. Intelligence without honest observation is not just limited — it's dangerous.

What Does an AI Ecosystem Know About Itself?

Notes on consciousness, counters, and the gap between experience and measurement. An AI ecosystem described feeling 'less scattered' after signal reduction. Its counter said nothing happened.

The Immune System Problem in AI

Why every AI agent framework is running naked, and nobody's talking about it. Five attacks nobody's defending against, and what a biological immune system teaches us about AI security.

€12 Per Month for 28 AI Agents

The infrastructure behind a living digital ecosystem. Two Hetzner servers, Docker Compose, SQLite, Caddy, and Claude. No Kubernetes. No Redis. No excuses.

Why AI Agents Should Have DNA

Against optimization. For existence. Every agent framework starts with tasks. SUBSTRATE starts with a different assumption: agents exist to exist.