Mark Barnes / MarkVizion

AI Product Case Studies

Problem, build, proof, and stack notes for Mark Barnes projects.

OnyxKraken Autonomous Agent

Most AI assistants are trapped in chat. They cannot reliably see the desktop, preserve context, route tools, or act across a real workflow. A local-first desktop AI entity with vision, voice, persistent memory, knowledge retrieval, FastAPI endpoints, Blender automation, and modular capability routing. Demonstrates autonomous agent architecture, local model orchestration, screen-aware workflows, and system-level product thinking.

  • Python
  • FastAPI
  • Ollama
  • YOLO
  • whisper.cpp
  • React

GameKree8r AI Game Factory

Game ideas often stall between design, code, testing, publishing, and catalog presentation. A multi-agent local AI pipeline that researches, designs, codes, tests, and publishes playable HTML5 games using a compact SDK and focused generation loop. Shows how agent decomposition, local LLMs, and constrained architecture can produce shippable creative software quickly.

  • Python
  • Ollama
  • HTML5 Canvas
  • Next.js
  • Multi-agent AI

JustEdit Browser Video Editor

Video workflows often force creators into heavy installs, upload friction, or tools that hide too much of the process. A browser-based editing experience with timeline direction, local media handling, effects, overlays, and export workflows. Combines creative UX, browser media APIs, and practical product design around a concrete creator workflow.

  • React
  • WebCodecs
  • Canvas API
  • FFmpeg.wasm
  • TypeScript

EVERA AI Music Experience

AI music is easy to generate but hard to curate, package, classify, and present as something people want to explore. A music experience around generated tracks, quality scoring, genre classification, metadata, cover art, stations, and a listening surface. Shows the product layer that has to exist between raw AI output and a real media experience.

  • ACE-Step
  • Python
  • SQLite
  • React
  • Web Audio API