9TEVE-O · AI Gardener · Developer · Founder, LAB AI & Tech Solutions
Building AI that can be questioned.
I have a habit of becoming interested in the parts of technology that everyone else walks past.
The awkward handoff. The missing document. The model that sounds certain when it shouldn't. The workflow that takes twelve steps because nobody stopped to ask why there were twelve steps.
So I pull things apart.
Sometimes it's a system. Sometimes it's a process. Sometimes it's three models, six documents, a half-finished idea and one person who knows where everything is. Then I start putting it back together.
My work moves between software development, AI research, document intelligence, retrieval, automation and creative technology. I came to it through operations, public safety, compliance, events and music, which taught me to notice what breaks quietly and what people actually do when the procedure says something else.
I call myself an AI Gardener because that's mostly what the work feels like: plant something, watch it grow, cut out what isn't useful, and keep the parts that work.
Five public repositories representing software, experiments and tools. A repository's presence here is not a claim that it's production-ready.
ASPRON · Safe Intake
A browser-based proof of concept for bounded information processing, safer intake, redaction and human-controlled review.
AI Policy & Terms Analyzer
Python tooling that extracts technical and operational signals from public policies, terms of service and privacy pages for structured human review.
Google Drive Organiser
A React and Express application exploring Google Workspace organisation, AI-assisted file management and productivity workflows.
Still Wild Garden
A quiet pixel garden designed to develop at its own pace. No tasks, scores, streaks or punishment for leaving. An experiment in more considerate digital experiences.
KinSync
A TypeScript family-coordination application scaffold with a Next.js frontend, Express API and separated supporting services. Still under development.
Some pieces are technical notes. Others are essays about people, music and technology. I try to separate what has been demonstrated from what merely sounds convincing.
- Ending Vibe-Based AI: Building Deterministic Evaluation Gates for Production LLMs · Technical note on explicit acceptance criteria and release decisions.
- 20 AI Instructions I Actually Use · Sep 2026 · Practical controls for reliable AI workflows.
- How to read an AI launch announcement · Jun 2026 · Distinguishing claims, testimonials and tested results.
- Technology Is Still Archaic Without Wisdom · May 2026 · When sophisticated technology still asks too much of people.
- We're Talking to Rectangles · May 2026 · Voice, making things and the strange experience of talking with AI.
- The Music Industry's Real Data Problem Is Time · Apr 2026 · Temporal reasoning and music information.
- Spotify’s AI Coding Shift Is Not Really About Code · May 2026 · AI infrastructure and the music industry's visibility into platforms.
- QLD AI Festival 2026: Is the music industry actually in the room? · May 2026 · Whose interests are represented in AI discussions?
- The Garden Can Wait · Jun 2026 · Music, memory and attention.
- I Thought I Was Building It for Her · Jul 2026 · Why making things matters, even when the first version is rough.
Read The Almost Useful Project on Substack
Question → Research → Prototype → Test → Prune → Verify → Document
I'm interested in the distance between something appearing to work and having enough evidence to trust it. That means examining sources, assumptions, failure cases, permissions and where a human needs to make the decision.
Not every problem needs another agent, framework or layer of automation. Sometimes the best improvement is removing something.
Alongside my independent work, I'm developing LAB AI & Tech Solutions around a practical question: how do we make complex technology genuinely useful to the people working with it?
My focus includes AI-assisted workflows, document intelligence, automation and software development, with clear boundaries and human judgement.
I like comparing notes with people who are curious enough to build something and careful enough to test it.
GitHub · The Almost Useful Project
Build things. Break things carefully. Learn in public. Keep what grows.