Tell us a bit about yourself and your business, and our experts will help get you sorted with the right solution for your needs.
Could AI agents help your organisation improve productivity and scale up more effectively? Yes, but not without plenty of human expertise and oversight, according to Datacom’s AI Solutions Lead Clint Francis and AI Practice Manager Angela Millward.
An AI agent uses artificial intelligence to autonomously accomplish tasks. It brings together information from large language models (LLMs), with internal company-specific data, to be able to deliver on complex goals without human intervention.
You might already be familiar with a standard AI chat interface, where you type in a question and the chat hands you back an answer. But an agentic workflow is quite different, explains Francis.
"It breaks down inputs into a series of steps: planning, tool use and reflection. It takes your request and decomposes it into smaller subtasks, then identifies a plan for how to proceed. The agent can then use the available tools, including mechanisms for accessing external information or triggering processes in the outside world. Once an output is generated, it reevaluates the result to ensure it lines up with the initial question, and, if it does, it will hand you back the outputs."
An AI agent is ideally suited to carry out the kind of repetitive or mundane tasks that most of us don’t enjoy or find difficult. Large-scale mathematical calculations, sifting through vast quantities of data or analysing research in an unbiased way, for instance.
“The RAG [retrieve, augment and generate] pattern allows the LLM to interface with an internal database,” says Francis. “It can search for topics like a researcher, then synthesise it and pick out the information that lines up with what the user asked for. For instance, you could ask it to correlate current sales numbers with lead generation and average closing time. The AI agent is making your data respond to conversational language.”
Call centres are already being transformed by AI agents, by providing customer responses that are accurate, contextualised and even delivered in the language of your choice. Universal real-time translation, for example, means you can speak in English on a Teams call to a client in China, and they hear your voice speaking in Mandarin.
“Or you might be running a one-person business, and you can set up an AI to be your assistant, receiving phone calls, answering questions and triaging, then sending you an email with all the details and a list of your actions,” Millward says. “It will then send a verification email to the caller to make sure it got everything correct. That’s an entire workflow we’ve built.”
These types of tasks can streamline operations, improve productivity and drive efficiency for organisations of all sizes. The use of agentic workflow in app modernisation is another example, where a job which might have traditionally taken two years is now being completed in a few months — or even days.
“The old method of legacy migration is not what I’d call fulfilling work,” says Francis. “We can now take the process down to 15 or 20% of the time taken previously. These are massive productivity gains. Instead of just trying to get to parity, people can now focus on value add-ons with their modernised app.”
This example highlights a growing recognition that when it comes to mission-critical IT infrastructure, "bigger" doesn't always mean "better."
AI agents might be autonomous, but that doesn’t mean humans are redundant. The ‘human in the loop’ is essential to the success of agentic workflow. The AI output must be checked and overseen, because no matter how good the technology is, it won’t be accountable for mistakes.
Consider a situation where legal firms hand discovery over to an AI agent, Francis suggests. It might provide some good results, which need to be reviewed by an experienced senior lawyer. In future, how will upcoming lawyers effectively review AI discovery if they’ve never done discovery training themselves?
“It can’t be blind trust. You need people to review the outputs if you really care about the impact on your business,” Francis says. “Agentic workflow isn’t replacing people, it is task orientated. Having a human in the loop is super important; if you completely rely on machines, that’s when you’ll run into issues. You need people who understand what’s happening.”
Millward says that business leaders introducing AI agents need to think about a range of governance aspects: policies, guardrails and ethics. Get it right, though, and this technology provides the chance to design a better future for everyone. With the population forecast to decline and age up, AI agents can fill gaps in skills and capability across a range of industries.
“Look at your own business and think beyond just replacing what you do today,” Millward says. “You have the opportunity to embrace AI agents and change your entire business. By designing and working through people’s lived experiences, we can control the future we create and design it to be what we want it to be. We can get computers working for us instead of the other way around.”
Artificial intelligence is transforming work and life, driving a fundamental shift. Datacom blends deep local expertise with global AI capabilities to deliver human-first AI solutions for enterprises and government.
Our focus is on building strong foundations that help organisations harness AI effectively to accelerate their journey and enhance — rather than disrupt — the workforce. Across Australia and New Zealand, our consultants support clients with governance, proof-of-concepts, AI tools and safe experimentation environments, empowering smarter decisions and greater impact.
