Roadmap
Two months to Doral. Eighteen to seize whatever opens.
By the G20 at Doral on December 14–15, we will have filed a 200-page Strategic Memo v3.0 with the administration and held our first hosted Consultations with delegates of key US AI powerbrokers. By March we grow from one person to four. Then we follow wherever the window opens, against the milestones below, with explicit review and stopping criteria.
Co-lead it, or be cornered by it
In one September week, 29 nations and the European Commission called for an international institution to control frontier AI. Xi told Trump that "AI must be kept under human control." Washington said no to both.
As we argue in an October 1st blog post, Cornered, Beijing could cheaply turn that gap into a UN-anchored treaty assembly, with America's chair left empty. Trump can still co-lead it — at Doral, or when his Super Intelligence Force reports around February 1.
Our launchpad
● The text: a 356-page Strategic Memo (v2.6), updated through v2.9 and built on 950+ articles, videos and papers; v3.0, about 200 pages, due November 20.
● The map: 150 pages of powerbroker profiles, and a tiered list of ~30 targets ready on day one.
● The channels: over 30 hours with key senior AI staff reporting directly to four of the twelve identified key US AI policy powerbrokers.
● The bench: 95+ people close to US AI powerbrokers engaged, plus 18 advisors and 26 contributors.
● The engine: a 270-paper library, a ~1,000-entry evidence base and a daily treaty brief.
All on about $140,000 and 2,100+ volunteer hours. Achievements →
Months 1–3 · To Doral
Key dates: Nov 3 midterms · Super Intelligence Dialogue, by November · Nov 18–19 APEC, Shenzhen · Dec 14–15 G20, Doral.
The Mission
● Memo v3.0 on file by Doral: a two-page brief for the President, an economics chapter, and the powerbroker profiles recast as a coalition map around seven AI predictions.
● Consultations, provisionally December 9 in Washington and 13 in Miami, then January 13 and February 10 in Washington, plus 20+ delegate sessions, with all four warm channels back to monthly contact.
● Three briefs: for the Super Intelligence Force, an incident rulebook for the Dialogue, and a co-lead brief for Doral.
● An Open Letter to Trump, plus a day-after kit for the next serious AI incident.
The Scaling
● Bridge: $60K now funds six months, through the Force's report.
● Two 18-month budgets:
○ Minimum, $950,000 (about $53K a month): a team of four, quarterly Consultations.
○ Ideal, $1,750,000 (about $97K a month): market salaries, Consultations every six weeks.
● Fractional COO: WorkStream Nonprofit, helped build both budgets; its operational role depends on the funding agreement.
● Three hires by March, onboarded by the COO:
○ a Head of Policy and co-founder in DC, for powerbroker relations;
○ a content lead, for memos, briefs and the economics chapter;
○ an AI-native chief of staff, who turns meetings into follow-through.
Months 4–18 · Wherever the window opens
What comes next depends on Doral, the January 13 and February 10 Consultations and the Force’s report. One thing is fixed: 80%+ of leadership time goes to the mission. Then:
● Trump moves: we surge, and v3.0 becomes a working draft for his envoy.
● Talks grind on: we widen the coalition, rung by rung.
● Others move first: we work to make America's seat a co-chair, not an empty chair.
● An incident hits: the day-after kit goes out within hours.
● Nothing moves: we reassess and redirect.
We will prioritize outputs useful across scenarios and reassess activities that are unused or risk harming other safety work.
Hold us to these numbers
● Month 3: bridge closed, v3.0 on file, first Consultations and 20+ delegate sessions held.
● Month 6: a budget funded, team in seat, a working group of 4+ delegates.
● Month 9 (the gate): 3 of 4 channels sustained, 6+ delegates, Plan v1. If the gate is missed, we document why and obtain external review before continuing, narrowing or redirecting the work.
● Month 18: an AI Diplomacy Preparedness Plan, with Track II to Beijing and a verification feasibility group running.
On the Minimum, we scale to the Ideal only after the gate.