Over the past few weeks, Octane Solutions ran AI roundtables with finance leaders in Perth and Melbourne. No decks, no slideware, no "AI 101." Instead, we sat down with CFOs and their teams, showed them real AI models solving real finance problems, and let the conversation go wherever it needed to go.
That last part mattered more than we expected. These sessions weren't a pitch delivered at finance executives — they were a conversation with them. And what came back was some of the most candid, useful feedback we've heard on where finance actually stands with AI today.
Rather than talking in the abstract, we walked through live AI models built around finance use cases — reconciliation, forecasting support, reporting automation, and more. Seeing the technology work against problems finance teams recognise immediately changed the tone of the room. It moved the conversation from "what could AI do" to "here's where this breaks in our environment, and here's where it doesn't."
Almost every finance leader in the room told us the same thing: they already have Copilot in place. And almost every one of them said the same follow-up — it's lifted individual productivity, but they haven't found a way to turn that into enterprise-wide impact. The tools are there. The organisational leverage isn't.
That gap — between individual convenience and enterprise ROI — became the throughline of both sessions, and it's a theme we'll come back to later in this series.
Nobody in the room debated whether AI belongs in finance anymore. That argument is over. The harder question — the one every CFO is actually wrestling with — is how to get from pilots and productivity tools to projects that deliver measurable, defensible return. That's where most finance functions are getting stuck, and it's the gap our workshops are built to close.
If there's one thing we'd tell any finance team starting this journey: don't start with technology, start with your own use cases. Bring your data governance and audit requirements into the room from day one, not after a pilot has already been built. And be honest about what "enterprise productivity" actually means for your organisation, versus what individual tools can already deliver.
In the next two posts in this series, we'll dig into the two issues that came up most: why finance-specific governance and audit requirements catch most AI pilots off guard, and why the cost of frontier models is becoming a real problem for finance to control — and what we think the path to real ROI looks like.
Join the conversation. We're running the next session of this roundtable in Sydney. If you're a finance leader trying to move past pilots and Copilot licences into real enterprise ROI, come and be part of the discussion. Register here to secure your seat.