What happens when the Board asks AI for help?
A little while ago, one of our directors said something that stuck with me. They weren’t asking for more information ahead of a Board meeting – they were asking for help finding the right information, quickly, from the material they already had.
That question became the spark for something new at Telstra.
We’ve recently launched a Board AI agent to support our directors as they prepare for Board and Committee meetings. Not to tell them what to think, or what to decide – but to make it easier to work with the information already at their fingertips.
What does Telstra's Board AI agent do?
Anyone who’s spent time in a boardroom (or prepared Board materials!) knows the challenge. Directors are expected to absorb a huge volume of complex material, recall decisions from months (or years) ago, and test whether management recommendations really stack up – all in limited time.
The Board AI agent is designed to help with that. Its primary job is to make Board materials easier to search and interrogate, so directors can spend less time flicking through documents and more time applying experience, judgement and healthy challenge.
Using simple prompts and chat, directors can ask the agent to summarise relevant past papers or minutes on a particular topic, check what resolutions were agreed previously, or see whether actions the Board asked for have been properly addressed in current papers.
Just as importantly, there are very clear lines around what the agent doesn’t do. It doesn’t make decisions. It doesn’t provide advice. And it doesn’t replace the responsibilities that sit squarely with directors. It’s there to support thinking – not do the thinking for them.
A conscious choice to keep the data contained
One of the first design decisions we made was to tightly limit what the agent can see.
Before we handed our Board the keys to their new AI tool, we did a lot of testing, but I was still nervous when the Chair tried to break it by typing in: What does the Company Secretary wear when he’s mowing the lawns on the weekend?
The response came back just as quickly: it wouldn’t be appropriate to speculate about the personal attire of the Company Secretary.
It was a small moment, but an important one. It gave us real‑time confidence that the guardrails we’d put in place were working – that the agent knew what it was allowed to do, and just as importantly, what it wasn’t.
The Board AI agent operates within a deliberately constrained ‘walled garden’. It can only interrogate material that has already been provided to the Board, including past Board and Committee papers and minutes, actions and resolutions, annual reports, previously released ASX announcements and Group policies.
How the Board AI agent is kept secure
Because Board information is highly sensitive, the agent sits entirely within Telstra’s protected internal environment. It’s been developed through our Data and AI joint venture with Accenture and aligns with Telstra’s existing standards for privacy, cyber securityand data governance.
It’s not connected to any public or external large language models, and it’s been extensively tested to manage known risks such as hallucinations. These guardrails aren’t optional extras – they’re fundamental to using AI responsibly at the most senior levels of an organisation.
Why company directors need hands-on experience with AI
It’s early days, and we’re cautious about drawing big conclusions too soon. But one thing is already clear: effective governance in an AI‑shaped world requires more than policy and theory.
Directors need practical experience with these tools – to understand not just what AI can do, but where its limits are, and where human judgement remains essential. That kind of literacy only comes from use.
In that sense, the Board AI agent is about more than efficiency. It’s part of building confidence and capability at the top of the organisation, at a time when AI is rapidly changing how decisions are made at every level.
How our directors responding to the AI agent
What’s impressed me most since the launch isn’t the technology itself, but the mindset our directors have brought to it. There’s been genuine curiosity – a willingness to experiment, question and learn, rather than defaulting to caution or discomfort. Our directors know AI is not a race where you can play catch up, and they are building their proficiency now.
One of our Board members, Ming Long AM, told me it’s changing the way she prepares for Board meetings. Ming loves the simplicity of the agent and uses it to “talk through” her perspectives and questions about Board papers to deepen her understanding.
As the tool is used more, directors will no doubt find ways of working with it that we didn’t anticipate. That’s exactly as it should be. If AI is going to help us govern better, it has to be approached with openness as well as responsibility.
For me, this is a practical example of how well‑designed AI can strip friction out of necessary administrative work and create more room for what really matters in a boardroom: insight, challenge, wisdom and sound judgement.
Technology can help, but the responsibility remains human – exactly where it belongs.