Overview
AWS tools for every level of expertise
Built-in governance helps to manage risk
The skillset to create an interconnected system
Discover more

Agentic AI on AWS: The tools with guardrails built in

AI learning tools visualisation inside a business office AgenticAIOnAWS_ArticleCoverImage_1920x600px@2x
  • AWS’s AI toolkit supports simple large language model (LLM) integration to enterprise-grade agents, as well as quantum options, with future enabling scalable, governance-ready deployments.
  • Built-in governance is embedded in the stack with Bedrock guardrails, tokenisation and policy integration providing risk mitigation and auditing, while AgentCore provides centralised oversight to define guardrails and monitor AI agents across their lifecycles.
  • Datacom’s rollout illustrates the journey from migration to AI-enabled operations, including AI app modernisation and potential transformation of .NET, Java and mainframe apps via AWS AI services, with holistic, cost-conscious tooling. 

The artificial intelligence (AI) stack from Amazon Web Services (AWS) is enabling the next phase of agentic AI adoption for organisations across Australia and New Zealand. 

The road from migration to real, transformational services can feel ambiguous, so it helps to get advice from someone who’s already putting the AWS AI tools to use. We talk to Paul Dunlop, Head of AWS at Datacom, about AWS’s latest AI capabilities, how Datacom has rolled out agentic AI across its operations and how governance is built in. 

At Datacom, directors are asking their senior technical leaders, “How are you embedding AI into your businesses practices?” 

“We’ve all had to take a step back and review a) what AI is changing and b) where we would best benefit from implementing these changes – not just internally but in ways that beneficially impact customer delivery as well.” says Dunlop. 

“It’s an interesting journey because we’ve moved into a framework of having virtual employees working alongside us, not just a chatbot customer agent, more as virtual sidekicks augmenting every staff member in their daily workflow. This has increased productivity giving rise to faster and more efficient customer outcomes.” 

aws-tools-for-every-level-of-expertise

AWS tools for every level of expertise

AWS has been developing its AI expertise for a decade, and its toolkit provides solutions for enterprises at every level. Whether your business is making its first foray into an internal LLM, or you have PhD-level expertise and want to build your AI from scratch, there’s something that will suit your requirements:  

  • End user: “For those of us who are not builders and prefer to use the pre-made solutions, AWS Quick Suite and Amazon Q Business provide a modified analytics service with a ChatGPT-style, end-user interface that simplifies accessing business data with AI-capable insights and automations. Additionally, Amazon QuickSight (not to be confused with Quick Suite) now embeds AI natively, and AI is baked into the AWS console giving you a new virtual team member able to analyse and query your apps and infrastructure easily.”
  • Integrating an LLM: “With help from Amazon Bedrock AgentCore and Strands Agents SDK, builders can integrate AI capabilities into pre-existing or new applications with no fuss. With even less fuss if they use Amazon Kiro and Kiro CLI where those tools can help to write and improve upon existing or greenfield code bases.”
  • Expert-level quantum computing: “If you’re a 'PhD' who’s keen on making your own LLM from the ground up, you can use the AWS Sagemaker service to make your own foundational models.”

“We’ve been helping customers at each of these levels,” says Dunlop. “Additionally, where customers have asked for help evolving their internal organisations to incorporate new processes or operational domains for AI we’ve assisted by providing enhanced shared-service foundations on AWS alongside involving our cloud rapid innovation development arm. If a customer needs an AI component, we can prototype a backend AI tool using our app modernisation arm, working backwards on each specific business problem our customers may have.” 

He adds that Datacom’s AI application modernisation services are an indicator of the future potential of AWS AI tools. Looking ahead, he anticipates that AWS AI transformation services, such as modernising .NET, Java and mainframe apps, will be crucial for helping customers leapfrog from legacy to modern stacks as part of their cloud migration journeys. 

built-in-governance-helps-to-manage-risk

Built-in governance helps to manage risk

AWS understands that agentic AI requires careful governance, and its expanded tooling now builds in agent oversight, rather than bolting it on. Bedrock guardrails, tokenisation and policy integration make compliance, risk mitigation and auditing a native part of the AI lifecycle.

“AgentCore helps enterprises customise and centralise their AI agents, and put guardrails up that keep it secure,” says Dunlop. “That’s another example of how AWS sees common trends and has good ideas about how to respond to the market, placing security at the heart of everything they do, never mind their capability to lead in the silicon market, creating AI-specific training chips, such as Trainium 2 and 3, which are purpose-built AI accelerators, along with Nova and Transform.” 

Dunlop and the AWS team at Datacom not only have the technical understanding to support customers with these tools, but they also have the experience of putting them into practice. 

“We provide customers with ideas and systems to transform their businesses as a result of how we’ve leveraged AI ourselves to transform Datacom. That includes migration, assessment and integrated agents that assist in repeatable tasks. Those agents lower the time it takes to do those tasks and allow us to address our internal operating models, uplifting our own organisational view on how we consume AI.”

the-skillset-to-create-an-interconnected-system

The skillset to create an interconnected system

Working out which tools are right for your enterprise isn’t always easy, and companies don’t always use a dedicated AWS stack. One strength of partnering with Datacom is that the team takes a holistic view of an enterprise’s stack.

“You might already be a consumer of AWS, or you might be agnostic and just want to find the best approach to get AI. We can use AWS or Microsoft Azure tools, and we have the skillset to create an interconnected, hybrid system based on your security requirements. Because AgentCore is open-source friendly, you can still observe all your agents from the AgentCore plane no matter where the LLM is running,” says Dunlop.

As an additional advantage, Datacom can often provide its customers with access to AWS services at a lower cost than they might get going direct. 

“We’re partnered very closely with AWS and LLM providers such as Anthropic,” says Dunlop. “We work to be as nimble as possible around the AI suite as it’s an ever-evolving and fast-changing landscape.”

Taken together, these capabilities, tools and guardrails underpin digital resilience for Australia and New Zealand organisations, enabling modernised apps, secure data flows and resilient operations in a changing AI landscape. 

Learn more about Datacom’s partnership with AWS.  

discover-more