I'm AI lead at Dragonfly Thinking — a strategic-intelligence firm whose clients include serious institutions. It's where I see the demand for something like Calyx most clearly, because it's where I feel the gap myself every day.

UAE government Commonwealth Bank — pilots Australian super funds — in discussion

This is real work, not slideware. We've built multi-agent strategic-intelligence systems that are in production today with serious clients — the hard, high-stakes end of agentic AI: faster, higher-quality decision-making under uncertainty. I'm the AI lead on it. The point is simple — the people building Calyx already ship agentic systems at that level.

Calyx itself is a separate, earlier-stage bet. We're exploring three distinct uses of it right now. None of that is contracted — it's exploratory and verbal at this stage — but the demand is real, and recurring.

Three uses, today
01

Demo environment

A reskinned Calyx as the environment for running live demos of Dragonfly's multi-agent strategic system — and the deep, interactive analyses it can generate.

02

The company OS

Everything Dragonfly runs on already lives in GitHub. The team is about to start working out of Calyx directly, against the shared company repo.

03

White-label

I have a verbal agreement with Dragonfly's CEO that the firm might white-label Calyx to deliver its agentic strategic-intelligence system to its own clients — Dragonfly's product, delivered through a Calyx-powered interface.

Dragonfly used to run its own platform and pivoted away from it. But its clients still want an interface to work with Dragonfly's system — and Dragonfly has no appetite to build and maintain one. We're actively being asked for this.

That last point is the one I keep coming back to. This is recurring, qualified demand from clients who are not small. A white-labeled Calyx is the natural way to meet it — and it would put Calyx in front of exactly the kind of high-stakes, security-conscious enterprise users it's built for. If this moves forward, it's enterprise validation close to day one.

And it's not only Dragonfly's clients. I also run Dragonfly's AI-fluency courses — teaching teams to actually work with agents — and I watch the same thing happen in every cohort: capable people hit the wall of doing this inside code editors and bent-out-of-shape note apps. The need Calyx addresses shows up in the room, repeatedly, in people who've never heard the pitch.

To be clear: none of the Calyx engagements are contracted yet. They're exploratory and verbal. But I'd rather show you a real, live signal with its real edges than a tidy logo slide.

← Back to the memo