In early September 2026 I reflected on the growing strategic uncertainty as AGI/ASI looms, and I outlined a balanced venture‑capital approach that keeps existing AI‑application bets while carving out a dedicated “post‑AGI” investment mandate.
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic.
Job Alert: Neothera is hiring an Engg Lead!
I decided to allocate 20‑30 % of fund capital to “defensive” post‑AGI bets targeting safety‑assurance, containment, and risk‑modeling startups.
I created a dedicated “Post‑AGI” investment mandate to capture upside in safety‑assurance infrastructure, containment platforms, and second‑order effect enablers.
I prioritized startups that design architectural isolation layers – walled‑away AI components with no direct API exposure – because they reduce systemic risk and are likely to attract regulatory support.
In September 2026 I shared a stream of X posts covering the pause and planned restart of the India‑2025 report, the launch of Rivet—a U.S. dating app backed by a $10.5 M seed round—and a speculative idea to license “The Speaking Tree” brand for vegetarian spiritual‑tourism hotels, among many other market insights and personal reflections.
I announced that Rivet will launch in the United States with a $10.5 M seed round, positioning the platform around a “Social Matching” model where real people help each other find love.
I noted that the India‑2025 report will be paused in 2026 and resume in early 2027 to ensure fresher data.
I reported that India’s gated‑community population is about 85 m residents, with per‑capita income equivalent to Russia.
I emphasized that in outbound GTM the priority should be improving TOFU rather than conversion improvements.
I shared that my AI‑assisted podcast workflow processed 7 podcasts + 3 articles to generate timeline summaries and cross‑transcript highlights.
I introduced the informal term Spoonicorn to describe a fallen unicorn.
I promoted my colleague’s new book Super Cool Startup, now available for pre‑order on Amazon.
I warned that AI‑driven existential risk could ultimately cease human‑level finance spam.