Active Projects

Omni Repo — Git Cleanup (Feb 16)

• Status: ✅ Done
• Merged t-598-local into live (10 commits: pipeline, skill discovery, web citations, dynasty renderer, send_photo)
• Deleted 57 stale branches, removed ~20 stale worktrees
• Remaining: live, ava, t-604 (active pipeline), usr/ben/kidcam
• Could not push to remote — root filesystem is mounted read-only, /var/git/omni.git unwritable. Ben may need to remount or push from another session.

Automated Dev-Verify-Ship Pipeline (t-603)

• Status: Code on live, not yet wired to agentd-spawned pi agents
• Priority: Tier 1 — highest leverage task. Once working, agents ship code autonomously.
• Dependency: t-475 (CI lint/test failures) should be fixed first.

LLM Newsreader (t-488)

• Status: Phase 2 in progress
• Priority: Tier 1 — systematizes technical-economic reading pipeline
• Ben planned to finish and start using it Monday Feb 16.

Samba File Share (t-604)

• Status: Task created, pipeline agent working on it
• NixOS config for SMB over Tailscale, read-only guest access to /home/ben/ava

Fund Model Dashboard

• Status: ✅ Built, at /home/ben/ava/fund-model/app.py (port 8235)
• Next: Ben to review and test

MSTR Preferreds Research

• Status: ✅ Complete at /home/ben/ava/research/mstr-preferreds/outline.md
• Found 5 preferreds: STRK, STRF, STRD, STRC, STRE

AI Infrastructure Bottleneck Research

• Status: ✅ Report at /home/ben/ava/research/ai-bottlenecks/report.md
• Next: Review, use to narrow investment thesis to specific under-priced bottleneck plays

Investment Thesis

• Status: Early stage. Legibility framework written into ~/fund/analysis/investment-thesis.md
• Core thesis: AI dominates legible work (Scott framing). Demand for intelligence unbounded → bottlenecks strained to physics level.
• Short candidates: Infosys, Wipro, Cognizant, Robert Half, Manpower (legible labor at scale)
• Long candidates: Power, cooling, networking, HBM infra plays
• Next: Review bottleneck report, pick 3-4 names for 10-K deep dives

Preparation Deadlines (dynasty planning)

• 2026: Find CPA for $2M+ portfolios. Start trust attorney search.
• 2027: Trust structure drafted. Estate plan v1.
• 2028-29: Build deal flow for business acquisition.
• 2030: Lia starts school — education planning decisions lock in.

Notes

• Pi agents can’t push to shared remote (read-only FS). Cherry-pick workflow works for recovering their commits.
• Agent secrets management punted — not the actual blocker for pi web search (they have env vars, just need tool instructions).
• Key priority order: Fix CI → build pipeline → finish newsreader. Pipeline is the force multiplier.