Carry what matters.
The next run inherits the goal, the decisions, and the prior proof. Distinct strands become context that can travel together.
Explore the workI build the memory, authority, and proof that keep ambitious systems together. Clinical software, data platforms, AI agents, game worlds.
Houston, TX · Cameroonian roots · A continuous practiceDrag to turn · tap a strand
Memory, authority, and execution are separate concerns. The green thread is what must survive their handoff.
Eight projects on one thread: each film’s own line carries the next, framed by the value it proves. Opens on my voice from Figma Config 2021; the thread ends by drawing my portrait. Score: “UBEAT V1” by Hope Atina.
One material. Different constraints.
Follow the thread as the work changes.
The next run inherits the goal, the decisions, and the prior proof. Distinct strands become context that can travel together.
Explore the workAgents act inside bounded permissions. High-consequence work waits for the right person. The boundary is part of the system.
Explore AlmaClaude, Codex, Cursor, and OpenClaw keep their native strengths. One operating context connects the work between them.
Explore OrgXA result should leave the next person something they can inspect: a baseline, a measured change, and an honest account of what failed.
Open the ledgerDrag to turn · tap a strand
Separate histories become a shared starting point. The signal is carried by the structure around it.
The system I keep building toward
Autonomy demos well. Carrying the goal, the decisions, and the evidence across tools and people is the hard part.
The next agent gets the goal, the constraints, and the prior proof in one go.
Claude, Codex, Cursor, and other specialists keep their native strengths.
Only the high-consequence calls wait for a person.
The result returns with enough evidence to improve what happens next.
The working ecosystem
Production engineering, independent products, research, and creative practice. Explore a discipline, then a tool to see how I use it.
14 highlights · 59 tools to explore
Compose stateful interfaces around the work: learning progress, clinical workflows, and evidence-led decisions.
See it in the workFlagship / OrgX
OrgX turns agent activity into work you can account for. Decisions, artifacts, and receipts that still make sense after the chat is gone.

The output, the decision path, the quality bar, and the human boundary live on one proof surface.

Trust shows up as what the agent is allowed to do next.
Range under constraint
Learning, research, games, clinical software, markets. Different pressure every time. Same question: does it hold together?
Make AI-delivered work accountable after the prompt disappears.
receipts / trust / continuity↗02Ship continuously where reliability, privacy, and auditability are inseparable.
HIPAA / rollout / adoption↗03Warn before runaway memory or low disk headroom takes the Mac down.
Guard / launchd / local / brew↗04Add organizational memory and accountable execution where agents already work.
state / streaming / proof↗05Turn a learner's curiosity into a course they can preview, study, and return to.
pathways / lessons / progress↗06Keep the experiment connected to the ideas and architecture that produced it.
papers / architecture / experiments↗07Make a place, a vehicle, and a moment of control feel like one world.
driving / physics / worldbuilding↗08Give a market signal an explanation and a review path before it becomes a decision.
explanations / paper / review↗Product taste is collaborative
Config 2021 · watch from 02:51 ↗I started Figma and Chill because talking about design wasn't enough. It became a global BIPOC design community that learned by making things together.
“I’m less about talk and more about action.”Hope Atina · Config 2021