Antares: From AI pilots to enterprise-scale impact

Q&A with Aaron Cunnington, Managing Director at Antares Solutions

1. Where are organisations struggling most when it comes to scaling AI, and how are you helping address those gaps?

We see two patterns. Some organisations are stuck in pilot purgatory — they've run proofs-of-concept around knowledge retrieval or content creation, but aren't sure which use case comes next, or which will really prove value. Others see AI's potential but hesitate, waiting to see what their peers do first. The breakthrough is realising AI isn't just for finding information or executing tasks — its real value is in challenging how we think. Imagine an advisor agent your people can ask anything of, grounded in your company's own knowledge and ways of working, helping staff learn and grow from the best of what the business already knows. The bigger prize is flipping the ratio: most people spend 70% of their time doing and 30% thinking. Hand the admin to AI and you reverse it — 30% doing, 70% solving bigger problems. That's where we help: finding the use cases that genuinely move the needle and delivering them on our Q Platform.

2. Many organisations have moved to the cloud but still struggle to operationalise AI. What's missing, and how do you help move from pilots to production?

Being in the cloud and being ready for AI aren't the same thing. Many organisations lifted-and-shifted workloads but left their data fragmented across systems — exactly what stalls AI in production. What's missing is the production scaffolding: a governed data layer to ground the AI, security and identity controls that satisfy risk teams, monitoring and cost management, and real integration into daily workflows. A pilot succeeds in controlled conditions; production demands reliability, auditability and support. We close that gap by treating AI like any other enterprise system — built on Azure AI Foundry with proper LLMOps, grounded in the customer's own data through Microsoft Fabric, and secured with Entra and Purview. The leap to production is less about the model than the operating environment around it.

3. What changes to architecture and data strategy are essential to support scalable AI?

Here's what most people miss: very little of this is new. A unified, governed data estate, clean data with proper lineage and access control, treating data as an asset with real owners rather than a by-product of applications — these are foundations every organisation should have had regardless of their AI ambitions. We've been doing this work for nearly 20 years, so we've seen it before. What's changed is the cost of not having it. AI forces the hand of any organisation that hasn't done the groundwork, because it exposes every silo and inconsistency immediately, and at a pace traditional reporting never did. The fix is familiar — for Microsoft customers, consolidating onto Fabric and OneLake, adding a semantic layer so AI reasons over business concepts rather than raw tables, and building security in through Entra and Purview from day one.

4. How are you helping organisations manage risk and build secure, responsible AI in cloud environments?

Let's be honest about what's happening: your people are already using AI to get more done, whether you've given it to them or not. We want to embrace that value, not pretend it isn't happening. The real choice is this — if you don't give people the capability to do more with AI, you either miss the productivity or push them external and introduce real risk. Shadow AI, where staff paste sensitive data into consumer tools, is exactly what that looks like. So we enable AI properly rather than lock it down: finding the highest-value use cases and delivering them in a governed way, with safety built into the solution and people trained to understand the risks and actually adopt it. Whether it's Copilot, Copilot Studio, Azure AI Foundry or our Q Platform, the goal is the same — get the most from the technology, safely. Done well, the controls aren't a handbrake; they're what gives risk teams the confidence to say yes.

5. Can you share a customer example that demonstrates impact, and what advice would you give CIOs looking to scale AI?

One example we're proud of is our work with NRMA. They adopted our Q Platform to bring secure, enterprise-grade AI to their people — grounded in their own data, governed centrally, and built into daily work rather than sitting in a sandbox. The result has been adoption that sticks and real productivity gains, because it was safe and genuinely useful from day one. My advice for CIOs is simple. Start with a business problem worth solving, not the technology. Bring security to the table early as a partner. Build internal capability so you're not perpetually dependent on vendors. And measure relentlessly: if you can't show value, you won't earn the mandate to scale.