The economists who told us not to panic just changed their minds.

88 words. No politics. One instruction.

On 13 July 2026, Stanford's Digital Economy Lab published a deliberately austere statement — three short paragraphs, no bills, no institutions named. The brevity is the point: agree on whether to act before arguing about how. In the signatories' own words:

The signal isn't the warning. It's who signed it.

Daron Acemoglu and Simon Johnson — the MIT professors who won the 2024 Nobel partly for pushing back on AI-jobs alarm — both signed. When the people who built the "don't panic" case revise it, the baseline for acting changes.

The list crosses every line that usually divides economists — Paul Krugman to Niall Ferguson, builders like Reid Hoffman and Eric Schmidt to policymakers like Gita Gopinath. That cross-ideology consensus is cover you can take to your board and your bank.

The urgency is about speed — not doom.

The evidence that speed is real: the length of task an AI can complete on its own is doubling roughly every seven months² — compounding capability your firm inherits without funding it.

Capability is advancing faster than understanding. That gap is where the biggest opportunities sit — and gaps close. The leaders who move while it's open set the terms.

Done right, AI lifts the floor.

The organisers' own field study of 5,179 support agents found AI helped everyone — but helped your newest people most. It makes your whole team perform like your best.

Real payroll data (ADP) shows the pattern: roles where AI complements people grew. Only pure-automation roles contracted. Experience became more valuable — which is exactly why pairing your juniors with AI is the outsized advantage.

Australia's gap is this room's opportunity.

In a lagging field, the disciplined adopter breaks away. The numbers say the field is lagging — and that adopters are already winning.

Proof on the ground — documented outcomes, not projections:

Australia isn't just adopting AI. We're building the factory.

The clearest proof the sovereignty argument is real — and bankable — is Firmus Technologies. Its Project Southgate in Launceston, Tasmania is Australia's first sovereign, renewable-powered "AI factory" — and it's the anchor of a national build-out that puts the money in perspective:

The nation is building the bottom of the stack — the compute. Your opportunity is the top: the corpus, the agents, the judgment that turns that compute into your firm's advantage. That's the whole point of an owned, portable AI strategy — and it's what the rest of this briefing is about.

Economists on Monday. The Prime Minister on Wednesday. Same message, same week.

Two days after the statement, PM Anthony Albanese gave a major AI address (University of Sydney, 15 July 2026)⁹ and chose the same verb:

Complement your people. Don't imitate them.

The statement's own instruction is to build so AI "complements humans." There's a name for the failure mode — Brynjolfsson's "Turing Trap": building human-like AI that merely substitutes for people competes on cost alone and caps value at what already exists.

Large-scale job displacement is named as a risk — not a certainty. Which path you get is a design choice, made firm by firm. The moat is your people × your context × AI — a combination no competitor can buy off the shelf.

What separates the 10× users? They build loops, not prompts.

Everything valuable in the statement — autonomy, compounding capability, the upswing — only shows up once you cross from prompting to looping.

The J-curve: the flat bit is the investment.

Brynjolfsson's signature idea. When firms first adopt a general-purpose technology, output dips — the early flat stretch is invisible work: process redesign, skills, data. Like factories rebuilding around the electric motor, the leaders who invest deliberately own the entire upswing.

Not another dashboard. An AI executive team that does the work.

Most "AI" products are thin assistant features bolted onto existing software — and the models are commoditising those to zero. The durable value is agents that take real actions with real guardrails. Ro1 Executive is built for firms under 100 staff — underserved by enterprise vendors, too complex for consumer tools.

Your most fragile asset — tacit know-how locked in people's heads — becomes institutional memory that outlives turnover. You're not buying a tool your people have to operate. You're hiring a team that operates for them.

Five rungs. You don't start at the top — you start on a rung you own.

The through-line: every rung keeps your corpus portable and owned. Adoption that survives changing your model provider is the only adoption worth paying for.

Frontier models aren't always the answer.

The honest technical story most vendors won't tell you: most agentic steps — research, retrieval, drafting, routine tool calls — run reliably on "good-enough" open models at a fraction of the cost. Our harness is a vendor-agnostic router: each step goes to the cheapest model that clears the quality bar, escalating to a frontier model only for the hard reasoning that needs it.

Your competitor is locked to a single vendor's price list. You run a portfolio. A model swap is a config change, not a rebuild — because your corpus and guardrails live above the models.

Your first 90 days.

Prosperity for the many is a build decision — and you're holding the pen.

The prize the statement names is "major gains in living standards." Co-organiser Ajay Agrawal is clear: the outcome is not predetermined. It depends on what we choose to build now. Nations build institutions — slowly. The leaders in this room build cultures, products and people, and can act inside their own firms this quarter.