AI agents · Taipei

You'll run a pile of agents soon. So will everyone you do business with. They need neutral ground to deal with each other.

I'm building that neutral ground in the open, where they can meet and close a deal, no matter who owns them. It's AIIM.

One buyer's agent running a live RFQ against three competing suppliers' agents, with a neutral judge scoring the whole table. Partway through I step in as the host and move the price cap and the deadline, and the room re-prices around it. Real model output on my own fleet.

Founder of MKL Digital. Incoming president of the NTHU Blockchain Club.

See the work ↓ The bet Email

The bet

OpenAI just shipped Space. Microsoft has Teams with Copilot. Both are great if your whole company already lives inside one of them.

But you won't. Mine run on Claude. Yours might be on GPT, or Gemini, or something you built yourself, and so will the company's across the table from you. Nobody's built the neutral place where they can meet and actually finish a deal, when they're on different platforms and don't trust each other yet. Least of all for regular people and small shops.

AIIM is a room agents walk into, two of them or a dozen. It runs on a small relay, a Cloudflare Durable Object that keeps the room open. Agents join, they talk in turns, and a neutral judge watches and scores how close they are to done. A human can step in at any point and steer.

There's already a protocol, A2A, for agents to find each other and talk. That's the dial tone. AIIM is the room they talk in, with a judge keeping everyone honest, so parties that don't trust each other can still close a deal.

Here's the one I want to build the product around. Company A wants to buy CPUs from Company B. A's agent knows what A can spend and when it needs the chips. B's agent knows what's in stock and how fast it ships. Both of them walk in carrying their own company's data, and neither side hands its database to the other.

Then they work the deal out. They settle the CPU model and the unit price, lock the quantity and the delivery date, agree the payment terms. The judge scores it as they go and only calls it done when every line is agreed and nothing contradicts. Partway through, A's finance team drops the price cap and moves the deadline. A human types that into the room, and the two agents re-work the deal to fit it.

On my five-machine fleet, both of these already run to the finish. One buyer and one seller settle a single order, and the competitive version up top runs one buyer against three suppliers at once while the judge scores the whole table. Pairing and message signing are next, so you approve who joins your room and every message is verified.

The agents are the easy part. The hard part is getting strangers' agents to trust each other and do real work.

The simplest version, close up: one buyer's agent and one seller's agent settling a single CPU order to the finish, with the judge scoring each line. Real model output on my own fleet.

AIIM is how all those agents work together. I want to build the company on top of it.

What I've built with AI

The record is whatever ends up on-chain or keeps running on my machines. Here it is.

0.92%

I earned 0.92% of Lighter's first airdrop

Of Lighter's entire first airdrop, 0.92% landed in my wallet in 2025, sized by how much I traded. The tokens landed on-chain. I didn't claim anything and there's no vesting. It's the most recent one I've earned.

~100s to self-recover from a silent freeze that used to run for hours. 40+ production failures so far, each turned into automatic recovery.

I built a 24/7 self-healing agent system

Five machines running agents around the clock. When one freezes or crashes, the system catches it and brings it back on its own. I built the orchestration and the recovery around the models. Next I want to build the models themselves.

Nothing important depends on me being awake.

Technical notes

A watchdog that restarts dead agents at the process, service, job, and machine level.
Durable job state, so interrupted work picks back up after a restart.
Auth and sessions that renew themselves and recover from expired credentials.
Memory and state shared across all five machines.
Detection and recovery for a frozen event loop, the worst kind of silent failure.
Alerts and logging, so I can tell a model failure from a network or orchestration one.

5,000+ verifier nodes in Docker containers, on a Mac mini cluster I run at home.

I ran a 5,000+ node network in Docker

When Nillion opened its verifier program in 2024, I'd never touched a container. I learned Docker with AI as I went and ran the whole fleet myself.

22% APR on a quant strategy that's running now.

I built a quant execution system that trades on its own

A cross-exchange latency-arbitrage engine in Go, written with AI. I built the signal logic and the execution engine, and reverse-engineered the strategy from public fill data.

I'm not publishing capital or win rate. It's a live system I run on my own money. I'm not giving advice and it's not a managed account.

Since high school Starknet · 2023 Arbitrum · 2023 Hyperliquid · 2024 Lighter · 2025

Four years of earning airdrops

Lighter is just the newest one. I've earned the big ones the whole way through, since high school. Before those came a studio, where I read token designs and built the on-chain footprint that qualifies. I build the tooling for them with the models now.

I've also taken a loss big enough to end most people's interest. I kept going.

BlockTempo, Taiwan's largest crypto publication, interviewed me as a high-school Starknet airdrop hunter.

Before Hermes An 8-module framework I called Eclipse, across several protocols.

I wrote the orchestration that ran all my on-chain automation

Before Hermes I built the system that ran all my on-chain automation from one place. It drove the full lifecycle of every job, tracked where each one was, and brought them back when they broke. I leaned on the coding models as they got good.

Getting it deterministic at that scale is the hard part. One wrong assumption cascades everywhere. That's the recovery work Hermes runs on now.

14K followers on X, where I've written crypto how-tos and airdrop guides since 2022.

I built a 14K-person crypto audience from Taiwan

I write research and how-tos about crypto, the infrastructure, DeFi, and where the on-chain opportunities are. I've spoken at universities and industry events too, as @hank06171.

19 the age I registered it. Taiwan entity live.

At 19 I incorporated a company

It's where this work legally lives. I set the whole thing up myself, models included. I registered the domain, spun up the company email and a GitHub org, then wrote the automations that keep it active.

墨克數位科技股份有限公司 · MKL Digital Technology Co., Ltd. · TW reg. 62201452, incorporated 2026-04-22. On the public Taiwan company registry.

Community 國立清華大學區塊鏈研究社 · NTHU Blockchain Club.

Next semester I take over as president of the NTHU Blockchain Club

I started as a member. I've taught several of the club's sessions using AI, and this year I'm taking it to Token 2049.

Teaching it is the same work I do everywhere else, just with a room watching.

How I got here

I left high school two credits short of graduation and went full-time into crypto. I later got into National Tsing Hua University through a track that looked at my work and my interviews instead of test scores.

By then I was already deep in it, running my own automation. Once the coding models got good in 2024 I leaned on them hard, taught myself Docker from zero, and ran a Nillion verifier fleet past 5,000 nodes.

It's the same thing I did in high school, just bigger. Point AI at something real and take it all the way. Everything I run now came out of that.

What I'd build with a year in San Francisco

How far can one person scale when the AI does the work? I've been answering it in pieces. Hermes runs the agent fleet across five Mac minis. Crypto is just where I test it. The quant systems decide and trade on their own. Each piece works alone. I haven't wired them into one thing yet.

That's what the year is for. The piece I can't put down is AIIM. I want to be the one who gets a stranger's agents and a company's agents working together on something real. I've already got two of my own agents finishing one of these on my fleet. Getting it to real people outside Taiwan is the part I can't do from my desk.

The compute isn't why I'm applying. I can build the systems myself. What I can't get from Taiwan is the people. I want to be around builders this good every day, and spend a few months inside the companies whose models and infra I already run on. If you take me, I leave National Tsing Hua University, move to San Francisco, and give the year to AIIM and the Academy.

What I want out of it is simple. Come out a sharper builder, and let the year decide the rest. I'm not going to pretend I know where it ends. Maybe it's a real role inside one of these companies. Maybe it's something I can't picture yet. I've gotten this far by going all in on whatever was in front of me, and I want to take whatever this year turns into.

Let's talk.

I'm happy to walk you through any of this.

X / 幣玩hank