These are selected public decision memos. Each memo keeps its own evidence date: the thesis is what we believed, the test is what it had to do, the deadline is when we decide, and the review is where we own the call. Omission means only that a public memo has not been published here; it does not mean a product is unregistered, retired, or a hobby.
current0 weekly activespre-launch · testflight · as of 2026-06-09
Thesis
Adults on GLP-1 therapy in India will adopt a clean dose-tracking tool that produces a doctor-ready PDF. Monetization comes from pharma B2B and telehealth referrals, not consumer subscriptions.
Test
1,000 active weekly trackers + 1 signed LOI with a pharma or telehealth partner.
Deadline
2026-09-30 — end of Q3.
Review
Under 200 weekly actives by deadline, OR no pharma/telehealth conversation past first call. Then: pivot or kill, no soft-landing.
status · legacy personal-team testflight build 26, live 2026-07-04 · internal beta · rebranded from myglpcare 2026-05-24memo last reviewed · 2026-06-09 · status line updated 2026-09-24
02Memo
Markets · India
Premium Chasers
heldredevelopment · decision paused
currentredevelopment heldREADME-only · no implementation, deployment, or release path · reviewed 2026-07-12
Archived thesis
Indian F&O retail traders want educational tools (payoffs, Greeks, paper trading) more than signals. We can build a paying education product without SEBI Research Analyst registration, which is the gate before any signal product.
Archived test
100 paying education-only users at ₹299–499/mo, with retention >60% at month 2. No signals required to clear the bar.
Former deadline
2026-12-31 — the archived beta memo's former end-of-Q4 deadline; it is not an active release commitment.
Archived review
If education-only retention is under 40% at month 2, or if we can't hit 100 paying without leaning on signals, the unit economics don't work here — kill, don't pivot into a SEBI-gated product.
status · previous implementation archived 2026-07-11 · README-only · Actions and releases disabledoriginal memo · 2026-06-09 · lifecycle reviewed 2026-07-12
03Memo
Cloud learning · global
Hands‑OnAI Labs
retiredretired 2026-09-20 · no decision pending
currentretiredretired from the portfolio 2026-09-20 · historical memo · original 2026-06-09
Archived thesis
Engineers preparing for Azure work want hands-on sandboxes with designed failure modes and explain-it-back checkpoints — not lecture videos. They'll pay if sandbox unit economics hold.
Archived test
Public launch by 2026-09-30, then 500 paying users/month on the platform within 6 months. Gross margin per active >50% after Azure pass-through.
Former deadline
2027-03-31 — the archived memo's six-months-post-launch decision point; the product was retired before it.
Archived review
Waitlist → paid conversion under 5%, OR per-user Azure burn breaks gross margin past 50%. Kill or pivot to a synthetic-sandbox model.
status · retired from the portfolio 2026-09-20 · renamed from drillkloud 2026-06original memo · 2026-06-09 · retired 2026-09-20
04Memo
Telugu · culture
ఆనాడు
—days to decision · 2026-12-31
currentlive · free · no adsshipped 2026-06-12 · as of 2026-07-01
Thesis
Telugu readers will build a daily habit around “on this day” history in their own language. The habit is the asset: free to read forever, with tasteful AdSense later — only once the habit is real. Ads before habit would kill it.
Test
1,000 daily readers arriving organically — search plus direct — sustained for 30 days, with zero paid acquisition.
Deadline
2026-12-31 — two full quarters after the 2026-06-12 launch.
Review
Under 200 daily readers by deadline: freeze investment. The nightly rebuild costs almost nothing, so the site stays up as a public archive — but no new curation hours and no ads on a site nobody reads.
status · live at aanaadu.com since 2026-06-12 · 762 pages · 96 timelines · rebuilt nightlylast reviewed · 2026-07-01
05Memo
Telugu · almanac
పంచాంగమనం
—days to decision · 2027-03-31
currenton holdinternal google play and testflight test builds only · hold decided 2026-09-23
Thesis
A Telugu panchangam app that shows its work and matches the printed almanac can earn daily trust, and Telugu users will open it every morning.
Test
Engine accuracy parity with published panchangams across a full Telugu year, then 500 weekly actives within one quarter of App Store launch.
Deadline
2027-03-31 — accuracy proven and the App Store launch call made by then.
Review
If the engine can't match published panchangams on tithi and nakshatra boundaries, the thesis fails at the root — fix the math or kill it. A wrong panchangam is worse than none.
status · internal testflight test builds since 2026-06-12 · google play internal testing since 2026-07-24 · on hold since 2026-09-23 · maintenance and fixes onlylast reviewed · 2026-09-23
Retired work / 07
Historical incarnations kept for the lesson
2026-04-17
OpenClaw — persistent agent fleet runtime that authored most of the early product code.
runtime retired
2026-04-17
Fairlypaid (earlier India/mobile incarnation) — the retired pay-transparency app used Supabase and TestFlight. The current registered Fairlypaid is the US DOL LCA browser product.
india thesis didn't fit
2026-04-17
moveitorthrowit — Stripe live, still serving on GCP standalone.
utility-pay thesis failed
2026-04-17
mission-control · neulern · openclaw-config — three cleared in one pass; Vidgen (earlier incarnation) was retired in the same pass.