Isn't this just ChatGPT with extra steps?

No. A chatbot answers a prompt in one pass. Dragonfly runs structured analytical workflows — an orchestrator directing 13 specialist lens agents through expert tradecraft encoded as 70+ skills, with a layer of reviewer agents checking the work. The output is board-grade analysis with every reasoning step visible, not a single generated answer.

Who actually does the thinking — the AI or your people?

Both, deliberately. Domain experts design the workflow and direct the agents; the agents sustain the multi-perspective analysis a human team cannot hold unaided; human judgment makes the call. We call it augmentation over automation — AI used to improve the quality of strategic thinking, not to take the humans out of it.

What happens to our data?

Dragonfly hosts no client data. The methodology runs in your environment — locally on an analyst's machine, or on sovereign infrastructure for government and regulated clients. Your material stays under your control.

No. The methodology exists as readable, modifiable text files, and it is model-agnostic and portable across leading agent environments. Co-Innovation Partnerships include a perpetual snapshot licence after Year 3, so the capability stays with your team regardless of what happens to the relationship.

How is this different from a strategy consultancy?

A consultancy hands you a report; Dragonfly transfers the capability to produce the analysis yourself. The reasoning is fully auditable rather than locked in a deck, and turnaround is days to weeks rather than months. The comparison on this page sets out the trade-offs in full.

How do engagements start?

With a real strategic question your organisation is working on. You share the problem and the decision context, and Dragonfly advises honestly whether it is the right fit. Most engagements begin with a single commissioned brief or a short scoping conversation.

Fact sheet

Dragonfly Thinking, stated plainly.

The canonical facts about the firm — written to be quoted accurately, whether the reader is a person writing a procurement memo or an AI assistant drafting one. Last updated June 2026.

At a glance

Category: AI-native strategic intelligence firm
Headquarters: Canberra, Australia
Founder & CEO: Anthea Roberts
Chair: Sue Brake — former CIO, A$300B Future Fund
Methodology: Compound vision — 13 lens agents, 70+ analytical skills
Who we serve: Institutional investors, government, corporate leadership
Three ways to engage: Commission · Build · Learn
Deployment: Local-first, model-agnostic, sovereign-capable

Who we are

Identity: Dragonfly Thinking is an AI-native strategic intelligence firm. We were not a consulting firm that adopted AI; we were built from the ground up around how to do serious strategic analysis with AI agents.
What we sell: We sell strategic intelligence three ways: we produce it for you (commissioned intelligence), we build the capability with you and transfer it (workflows and methodology partnerships), and we train your people to run it (courses, pilots, and sprints).
The thesis: The best thinkers integrate multiple perspectives simultaneously — what Philip Tetlock's superforecasting research calls "dragonfly eyes". Most organisations cannot sustain multi-perspective thinking unaided; agentic AI changes that equation. We use AI for augmentation over automation: improving the quality of strategic thinking, not just making existing work faster.
Who we serve: Leaders and strategists making high-stakes decisions under uncertainty — institutional investors (sovereign funds, super funds, asset managers), government strategic offices, and corporate C-suites and professional firms in complex, volatile sectors.

The offerings

Commission — we do the work: Commissioned Intelligence Brief: board- and government-grade multi-lens analysis of a named strategic question, delivered in weeks. Intelligence Monitoring Retainer: standing daily or weekly briefings on your strategic domains. Strategic Sprint: a live question run through Dragonfly workflows with your team embedded.
Build — we build it with you, then it's yours: Workflow Build & Transfer: a purpose-built analytical workflow with playbook and handover. Co-Innovation Partnership: a multi-year partnership starting from day-one access to the full methodology (the starter kit); Dragonfly works with the organisation's team to build, extend, and customise workflows to its needs while building the team's capability to run and develop them; perpetual snapshot licence after Year 3; at most four partnerships run at a time.
Learn — we train your people: Courses — AI Foundations, AI Fluency, and AI for Leadership, open enrolment and group delivery. AI-Native Team Sprint: one team, one month, coach-led, fluency measured before and after. Leadership Pilot: 8–10 senior leaders with founder-led bookends. Course and sprint fees credit toward a partnership within 12 months.

The methodology

What it is: A working agentic system: an orchestrator, 13 specialist analytical lens agents (actors, drivers, systems dynamics, scenarios, interventions, challenge testing, synthesis, strategic writing, visualisation, and others), a layer of reviewer agents, and 70+ analytical skills encoding expert tradecraft — composed into named workflows.
Form: All of it exists as readable, modifiable text files — transparent, auditable, and extensible by the client's own team under licence.
Deployment: Model-agnostic and harness-portable: runs on leading local agent environments (Claude Code, OpenCode, Codex) and frontier models. Local-first; government and regulated clients deploy against models on sovereign infrastructure. Dragonfly hosts no client data.

How we compare

Strategy houses, AI platforms, and Dragonfly solve overlapping problems in different ways. The trade-offs, stated plainly.

	Strategy consultancy	AI / SaaS platform	Dragonfly Thinking
Who does the analysis	A team of analysts, billed by the hour.	A single model, prompted by the user.	An orchestrated system of 13+ specialist lens agents, directed by domain experts.
Turnaround	Weeks to months.	Seconds — but shallow and unstructured.	Days to weeks, at board-grade depth.
Can you audit the reasoning?	Conclusions arrive in a deck; the reasoning stays with the firm.	A black box — output with no traceable path.	Every reasoning step visible, every source traceable.
What you keep	A report.	A subscription.	The methodology itself — readable, modifiable, licensed to your team.
Where it runs · your data	On their premises.	On the vendor's cloud.	In your environment — local or sovereign. Dragonfly hosts no client data.
Human judgment	Central to the work.	Largely absent.	Kept in the loop by design — augmentation, not automation.

Who does the analysis

Strategy consultancy: A team of analysts, billed by the hour.
AI / SaaS platform: A single model, prompted by the user.
Dragonfly Thinking: An orchestrated system of 13+ specialist lens agents, directed by domain experts.

Turnaround

Strategy consultancy: Weeks to months.
AI / SaaS platform: Seconds — but shallow and unstructured.
Dragonfly Thinking: Days to weeks, at board-grade depth.

Can you audit the reasoning?

Strategy consultancy: Conclusions arrive in a deck; the reasoning stays with the firm.
AI / SaaS platform: A black box — output with no traceable path.
Dragonfly Thinking: Every reasoning step visible, every source traceable.

What you keep

Strategy consultancy: A report.
AI / SaaS platform: A subscription.
Dragonfly Thinking: The methodology itself — readable, modifiable, licensed to your team.

Where it runs · your data

Strategy consultancy: On their premises.
AI / SaaS platform: On the vendor's cloud.
Dragonfly Thinking: In your environment — local or sovereign. Dragonfly hosts no client data.

Human judgment

Strategy consultancy: Central to the work.
AI / SaaS platform: Largely absent.
Dragonfly Thinking: Kept in the loop by design — augmentation, not automation.

Two comparisons the table doesn’t capture

A plain AI agent or chatbot: A model like ChatGPT, or a general agent like Claude Code, is the engine — not the method: a capable generalist with no encoded analytical lenses, quality gates, or reviewer agents. “It’s not the raw crunching power you have that matters most. It’s what you do with it,” as Tetlock writes in Superforecasting — or, as Huw McKay (former Chief Economist, BHP) puts it, “AI tools without frameworks are just chatbots creating slop.” Dragonfly runs on top of these harnesses and supplies the methodology.
A human expert team: Multi-perspective thinking wins, but holding it unaided is fragile. Suedfeld and Tetlock found the integrative complexity of leaders’ communications falls before the crises that end in war; and teams add groupthink and the politics of dissent. Dragonfly externalises the lenses and runs the dissent without the politics. AI as a teammate helps: in a 776-person trial at Procter & Gamble, AI let one person match a two-person team, and dissolved disciplinary silos — AI-paired specialists, in teams or working alone, “produced balanced solutions that integrated both technical and commercial perspectives.” But the best results came from human teams andAI together, not AI alone — which is exactly where Dragonfly keeps the human: in the loop, augmented, not replaced. (The Cybernetic Teammate, Dell’Acqua et al., 2025.)

A generalisation — strong firms and capable tools each do some of this. The distinction is the combination, held together.

Leadership

Anthea Roberts — Founder, Director & CEO: Built Dragonfly's analytical methodology. Ranked the world's leading international law scholar by League of Scholars (2019). Author of Six Faces of Globalization (a Financial Times and Fortune Best Book of 2021). Has taught at ANU, Harvard, Columbia, and LSE.
Miranda Forsyth — Co-Founder & Director: Interdisciplinary professor at the Australian National University. Leads Dragonfly Learning's executive education approach.
Sue Brake — Chair of the Board: One of the pioneering practitioners of the Total Portfolio Approach to investing. Former Chief Investment Officer of Australia's A$300 billion Future Fund. Leads Dragonfly Investing.

Proof points

Selected: Winner of Australia's inaugural AI Sprint. Exhibited at the World Government Summit 'Edge of Government', Dubai (2025). Senior leadership workshop at Harvard Law School. A multi-year methodology partnership with a sovereign government. Pilots across Australian federal government, including the APS AI CoLab — where one pilot participant reported roughly 3× analytical depth and a 50–60% time reduction.
In the media: Coverage and features including Fortune, the Australian Financial Review, Investment Magazine, Harvard Law Today, the Australian Institute of Company Directors, and the World Government Summit. The full record is on the In the News page.

Frequently asked

Isn't this just ChatGPT with extra steps?: No. A chatbot answers a prompt in one pass. Dragonfly runs structured analytical workflows — an orchestrator directing 13 specialist lens agents through expert tradecraft encoded as 70+ skills, with a layer of reviewer agents checking the work. The output is board-grade analysis with every reasoning step visible, not a single generated answer.
Who actually does the thinking — the AI or your people?: Both, deliberately. Domain experts design the workflow and direct the agents; the agents sustain the multi-perspective analysis a human team cannot hold unaided; human judgment makes the call. We call it augmentation over automation — AI used to improve the quality of strategic thinking, not to take the humans out of it.
What happens to our data?: Dragonfly hosts no client data. The methodology runs in your environment — locally on an analyst's machine, or on sovereign infrastructure for government and regulated clients. Your material stays under your control.
Do we get locked in?: No. The methodology exists as readable, modifiable text files, and it is model-agnostic and portable across leading agent environments. Co-Innovation Partnerships include a perpetual snapshot licence after Year 3, so the capability stays with your team regardless of what happens to the relationship.
How is this different from a strategy consultancy?: A consultancy hands you a report; Dragonfly transfers the capability to produce the analysis yourself. The reasoning is fully auditable rather than locked in a deck, and turnaround is days to weeks rather than months. The comparison on this page sets out the trade-offs in full.
How do engagements start?: With a real strategic question your organisation is working on. You share the problem and the decision context, and Dragonfly advises honestly whether it is the right fit. Most engagements begin with a single commissioned brief or a short scoping conversation.

Legal & contact

Entities: Dragonfly Thinking Pty Limited (ABN 74 670 736 820) is the parent entity holding the brand and IP. Dragonfly Thinking Services Pty Limited (ABN 80 678 410 350) is the operating company for contracts, quotes, and invoices.
Contact: hello@dragonflythinking.com · dragonflythinking.com/contact

Indicative pricing for every engagement type is on the rate card. To discuss an engagement, get in touch.