I like to put supercomputers in people's garages and give everyone an agent they actually own. 🤫🐶⚡ Founder & CEO of 🤫 hussh; steward of Alphabets 27. I think the next era belongs to people who own their compute, their data, and the agent that acts for them — by consent, on open standards.
That's the whole mental model. A garage owner buys good machines, keeps them well-maintained and well-stocked, puts them to honest work, and earns by serving the neighborhood. Do that with supercomputers — millions of them, in garages and warehouses, on Starlink, owned by ordinary people — and you get a distributed edge grid that gives everyone extreme computing at the lowest cost per watt, while strengthening their own communities. That's what 🤫 hussh is building, and why.
Two things, one thesis. At 🤫 hussh, I'm building personal agent infrastructure people own — Agent One, the Puppy One supercomputer, and the open consent protocol PCHP. With Alphabets 27, I steward a free-cash-flow growth portfolio for the people I'm accountable to. The thesis underneath both: own the durable, compounding things — your compute, your data, your agent, the best businesses — and do it the right way.
Personal agent infrastructure owned by the user — 🤫 Agent One, the 🤫 Puppy One supercomputer, the consent protocol PCHP, and a shared wiki/MCP memory for humans and agents. HushOne, Inc. & Hushh Technologies Corporation.
An evergreen, AI-first business partnership — a free-cash-flow growth & innovation portfolio for family, limited partners, and trusted operators and advisors.
Product leadership across AI/ML infrastructure and developer platforms — the cloud GPU/TPU and developer-ecosystem world.
Enterprise machine-learning product leadership.
Machine-learning products for the enterprise.
The roles that shaped a builder-operator's taste for clarity, trust, and infrastructure.
Department of Computer Science, with a minor in Mathematics, Management & Finance. A lifelong love of computers, math, physics, and programming — including quick-and-dirty games built for the pure joy of keeping his brain engaged.
Public context, assembled from public profiles and company materials by my own 🤫 Agent One.
The personal AI agent you own — it works from trusted context, memory, and consent, not an empty chat.
Open →🐶Your personally-owned supercomputer, working for you 24/7 — the edge of a distributed grid.
Open →🏷️The future companion designed with love for the dogs and relationships we cherish.
Open →🤝The Personal Consent Handshake Protocol — the HTTPS of the agentic era, published openly as RFC-001.
Open →🏛️An open, transparent coalition so everyone gets an agent they own, on a governed open protocol.
Open →📜A public, sourced map of the laws and institutions behind data ownership — and how to plug in.
Open →The best way to grow the field is to teach it in the open — in the spirit of the people I learn from. Here's what I keep adding to.
How to build a thriving open-source community by writing small, modular, self-contained code anyone can yoink — aligned to Andrej Karpathy's public teaching.
Read →🛠️Vibe coding, Software 2.0, the LLM OS, evals & taste — the simplest path for new builders, made open.
Read →📚The most important getting-started guides, tagged and organized by interest so anyone can find their way in.
Read →With admiration for Andrej Karpathy's public teaching — admiration of public work, not an affiliation or endorsement. See the people I look up to.
The future is a distributed edge grid owned by ordinary people — supercomputers in garages and warehouses, on Starlink, placed where cooling is abundant, power is cheap, and the sun keeps shining. Keep the garage well-stocked and well-maintained.
Prefer code anyone could lift without importing your whole world. Build a eukaryotic backbone only where complexity truly demands it. Maximize bacterial DNA.
Personal AI should not become another platform that extracts context from people. It should be owned, permissioned, and useful on the user's terms.
We do whatever it takes to win — in the open, on the merits, honestly. The pattern is to be everything but a crook.
Own the world's best businesses by absolute free-cash-flow and net-income growth into shareholder equity. As equity rises, risk falls — and keeping risk low is the goal.
Document every release publicly, run a Rude FAQ that tells the blunt truth, and keep the public story mapped to real, legible code.
This page lives at my hussh spaceID: manishussh — the public name I share to say who I am to the world (and to the stars at large). It comes free with my 🤫 One: every person who registers gets a default spaceID, and mine took the tag hussh. My Private Agent One's default 🤫 One ID is hussh, which becomes my 🤫 One Gold ID — the consent-first card that proves who I am without handing over my phone, name, documents, or card numbers.
And yes — this whole page is maintained by my own 🤫 Agent One. That's the point: a personal agent that understands the person it serves, keeps a durable model of who they are, and publishes exactly the slice they choose, by consent. The Personal World Model, made public by choice.
Happiness first: I've stayed happy-go-lucky by keeping my brain engaged — and my family is the center of it all. You can learn more about my family, by my consent, by subscribing in my 🤫 profile.
A spaceID, an agent that works for you, and a personal site like this one — they come free with your 🤫 One. Own your compute, your data, and the agent that acts for you.
One is a product of Hushh Technologies Corporation (brand: 🤫 “hussh”), an independent company. One runs on third-party silicon, systems, and cloud; all company names are used solely to describe the platforms on which One software runs. Hushh Technologies is not affiliated with, endorsed by, sponsored by, or partnered with any company named.