← Long Game
Law × AI × Crypto × Gamification · H2 2026v3.1 · 2026-07-20

My Builder Dashboard

A tech attorney starting with zero coding, using AI to build open-source tools at the intersection of law and crypto.

Why it matters: the digital noise in law and emerging tech is deafening. Building in public is how I cut through it — prove what's possible with AI, and find the co-builders and teams who want this space safer and more efficient.

The backstory: I failed a few times at Git and Python. AI changed the game. This site is my live playground, where I translate legal frameworks into executable web3 protocols.

Building & learning now, alongside my 12-year project PIF12:

  • AI & CS tech fundamentals
  • Blockchain Law RPG
  • CLO Cockpit
0% · 0 / 0
Your browser is blocking local storage, so ticks won't persist.
Now · Phase 0 · one next action

Loading…

Deadline inside the deadline: everything public finishes by mid-December. Anything still half-done then is, for every practical purpose, invisible.

The week: twelve hours, two lanes

Hours per week: courses left, projects right.

7 hrs
Learning — courses, protected
4 hrs
Building — projects, capped
Aug – Sep · foundations
7 hrs learning · 4 hrs building, a week
Fundamentals can't be crammed — they go first.
Oct – Dec · shipping
5 hrs learning · 6 hrs building, a week
Flips when the artifacts need finishing.
Learning 140h
Building 120h

~23 weeks · 260h. The twelfth hour is the queue slot — short lessons between other things. 12 is the target; 8 is the floor. A heavy client week cuts scope inside a project, never a whole lane.

Learning blocks are booked first; building gets what's left — that ordering is the whole defence. Every session ends with code I wrote or verified, not watched. Watching is not a session.

The five monthsWhat lands when

Aug — Python fundamentals; reverse-engineer the publish.py and GitHub Action already running this site. Building: the cockpit's connective layer.

Sep – Oct — Prompts, the Claude API, RAG, agents, evals, and the AI-governance write-up. Building: the game ships publicly.

Nov — On-chain data and a Solidity on-ramp. Building: the analytics dashboard, and both cornerstones close out.

Dec — Everything public finishes by mid-December. After that: write-ups, the portfolio page, the H1 2027 plan.

The guardrailThe illusion of the vibe-coding builder

Directing an AI to build produces real artifacts and no transferable capability — the sense that I can build is illusory, and the bill arrives the moment I have to verify, maintain or extend it on my own.

The tell: if I can't explain last session's code in plain language, that was consumption, not building.

Version historyWhat changed and why

v1
The first planCut the Stanford AI courses — wrong layer; they teach making AI, the goal is using it. Added the missing tracks: applied AI, smart contracts, on-chain data. Four phases, each gated.
v2
Widened the materialsAny format competes — one best fit plus one backup per module. Added the signal layer: proof nobody sees is half-finished.
v3
Two cornerstones and a budgetAdded the real projects, the two-lane budget, and the capability-debt rules.
v3.1
Half-year scope; one plan, one progress barRe-scoped to H2 2026 at ~12 hrs/week, deep crypto deferred to H1 2027. Header rewritten as a Smart Brevity brief and one word in five cut page-wide. Four tabs became two: learning and projects now share a single plan and a single progress bar. Dropped the standalone benchmark anchor — it is a by-product of the game, not a market I claim. Builder dashboard online.
How to read this The learning lane is protected; the gate tells me borrowed capability has become mine. Advance on the gate, not on hours logged. Projects sit in the same list and the same progress bar — the split was never two plans.

Projects

Two things I am building because they are useful and fun — if you want in, email me at projects@jasonjlai.net.

On radar

Live rather than scheduled — what I am tracking now, and what the next plan will most likely be built from. A fresh one gets written in December.

The blockchain market, up close and at a distance. Up close: real-world asset tokenization, and how institutions are actually entering rather than how they say they will. At a distance: blockchain application within and outside finance, with and without AI.

Ready once the fundamentals are in:

  • Solidity + Foundry in depth — Cyfrin Updraft, ~40h. The on-ramp is this year; the deep water follows.
  • Ethernaut — security intuition, which pays off once the basics hold.
  • ERC-3643 contribution — earned once I can read the standard in its own language.
  • AI-native rigor (CS146S) — richer after a year of directing agents daily.
  • The flagship — an agent producing a compliance read from on-chain data.
  • Also queued — the cockpit’s v2 collaboration layer, the game’s second content batch, the full Speedrun Ethereum track.
Disclaimer This is my website representing my view only, not my affiliated entities. All information is for informational purpose only. No specific legal, medical, tax, investment advice is rendered here. Seek your own professional advice. The content of this post is provided “as is;” and no representations are made that the content is error-free or up-to-date. Thus, please do your own research and take full responsibility for the consequences if you rely on any information here.