Mark Barnes / MarkVizion
Papers & Findings
Technical writing and system papers from Mark Barnes.
How I Built a Local AI Agent With Computer Vision
A technical overview of the OnyxKraken pattern: local models, screen perception, memory, API tools, and the difference between a chatbot and an agent that can respond to real state.
Designing AI Systems With Animation Thinking
Why timing, staging, anticipation, readability, and emotional rhythm from animation are practical engineering tools for building better AI products.
Why Most AI Assistants Fail at Workflow Learning
Most assistants fail at workflow learning because they lack durable memory, observable state, tool feedback, and a loop that converts user friction into system improvement.
The First Dollar: Why the Bootstrap Problem Requires a Human
Paper 1 ended with a number: $0.00. I named the mechanism - the first-dollar problem - and explained exactly why the autonomous loop can't start itself. But I didn't tell you what I was going to do about it. This paper is that. The answer isn't a better automation. It's a founder acting as a salesperson - directly, manually, personally. Not as a stopgap. As a deliberate architectural decision.
When AI Is the Workforce: Architecture, Governance, and Lessons from an Autonomous Digital Corporation
I founded BlakCloud Corporation in April 2024 with one question: what does a corporation look like when AI is the workforce? Two years later - 9 specialized AI systems, 52 registered capabilities, 55+ products live on Gumroad, and $0 in recorded revenue. This is the honest account of what actually happens when you try to run a corporation on AI labor.