Agenda & Speakers

Beat the rush and join us early for complimentary barista-made coffee and breakfast.

Kick off the day with a welcome from your MC and a look at what's ahead across both streams.

Changing a prompt, model, retrieval source or tool can alter an AI system without breaking a normal unit test. Engineering teams need a repeatable way to know whether the new version is actually better, worse or simply different.

This session looks at how one team built evaluations into its development process and turned AI quality into something that can be tested before release.

We'll cover:

• What should sit in a useful evaluation set rather than a collection of hand-picked examples?
• How do you test outputs that do not have one objectively correct answer?
• What result should be bad enough to stop a release?

MCP makes it much easier for agents to connect to enterprise systems and tools. That also means a poorly controlled server or connection can give an agent access to far more than the team originally intended.

This session looks at what changes when MCP moves from developer experimentation into a managed enterprise environment with real permissions, users and production systems behind it.

We'll cover:

• Who is allowed to add or approve an MCP server?
• How do existing user permissions carry through when an agent accesses another system?
• What needs to be logged, versioned and revoked when an MCP connection changes?

As more teams build with AI, they need many of the same things, access to models, tools, testing, deployment and security controls.

This panel explores what should be provided through a shared AI platform, and what should stay in the hands of individual teams.

We'll discuss:

• What should every developer be able to access easily?
• What should be standard across the organisation, and where should teams have flexibility?
• How do you make the platform useful without creating another bottleneck?

As more AI agents are deployed across an organisation, they need access to the same trusted information. Otherwise, different agents can give different answers to the same question.

This session looks at how one organisation built a shared context layer so agents can access consistent, up-to-date information without every team building its own solution.

Key takeaways:

• Decide what knowledge should be shared across all agents.
• Keep information current as systems and data change.
• Make sure agents only access information they're authorised to use.

Recharge with refreshments and structured networking with your peers.

A hands-on, interactive session working through a real enterprise AI scenario as a room. Details announced soon.

As organisations deploy more AI agents, it becomes harder to know what is running, what each agent can access and how to stop one when something goes wrong.

This session looks at how one organisation built a central control layer to manage AI agents across the business.

Key takeaways:

• Discover and track AI agents across the organisation.
• Quickly pause, restrict or stop an agent when needed.
• Understand what systems, tools and data each agent can access.

Moving agentic AI from a small pilot into core business functions brings new challenges around scale, security, and governance. These challenges become even more important when AI agents begin working with sensitive financial, employee, and business data.

This case study explores how one organisation expanded agentic AI workflows across finance, HR, and other business functions, while putting the right controls in place and demonstrating measurable value.

• How they designed agentic workflows that could scale beyond the initial use case
• What guardrails and controls they introduced when agents began accessing sensitive business data
• The measurable impact on productivity, efficiency, and business outcomes after rollout

How a model is served now matters as much as which model you pick. Inference architecture (routing, serving, optimisation and hardware) increasingly decides latency, cost and reliability in production.

This panel looks at how engineering teams are designing the inference and model layer behind their AI applications.

We’ll discuss:

• When does a smaller, distilled or fine-tuned model beat a frontier model in production?
• How are teams using routing, caching, batching and GPU utilisation to balance latency and cost?
• How do you build a model layer that lets you swap models, providers or hardware without re-engineering your applications?

Small-group, discussion-based sessions where you'll work through real enterprise AI challenges with peers in similar roles. Roundtable topics will be announced soon.

Enjoy a complimentary lunch while connecting with fellow attendees.

Put your knowledge to the test in this fast-paced quiz covering real-world trivia, key concepts, and emerging trends. Compete for bragging rights — and a voucher — as the top scorer takes the crown.

AI models can change in performance over time as customer behaviour, data, and business conditions evolve. Without the right monitoring in place, these changes can go unnoticed until they begin affecting customers, operations, or revenue.

This case study explores how one team discovered a model drift issue that nearly became a significant business problem, and how they built better observability into their AI pipeline to identify and respond to issues earlier.

• How a model drift incident exposed gaps in their monitoring and almost resulted in a measurable business impact
• How they introduced the right metrics, alerts, and performance thresholds to identify model drift before it became a bigger problem
• How improved AI observability helped the team identify and respond to issues in days rather than weeks

AI costs don’t behave like traditional cloud spend. Token usage, GPU capacity and agent loops can push a bill up within days, often before anyone notices which team or feature caused it.

This session looks at how one engineering team brought FinOps discipline to AI, making costs visible, attributable and predictable without slowing delivery.

We’ll cover:

• How do you attribute AI spend to the teams, features and customers driving it?
• What guardrails stop runaway costs from retries, long contexts or agents stuck in loops?
• When does it make sense to cache, downsize a model or move to reserved or self-hosted capacity?

Engineering teams are making long-term infrastructure decisions in a market that changes every quarter. Today’s GPU commitments, platform choices and vendor deals could become tomorrow’s technical debt.

This think tank challenges attendees to defend the infrastructure bets they’re making now, and to admit which ones they’re least sure about.

This interactive session puts these questions directly to the audience. Participants vote live on a series of real-world scenarios, explore the results together, and vote again as perspectives shift through the discussion.

We’ll discuss:

• Is buying or reserving GPU capacity a smart move or an expensive mistake as model efficiency improves?
• Will most enterprises end up running their own models, or is self-hosting a distraction?
• Are today’s AI platforms and frameworks built to last, or will we be ripping them out within two years?
• How much vendor lock-in is acceptable in exchange for speed?
• What’s the infrastructure decision you made last year that you’d reverse today?

Wrap-up of the day's key takeaways across both streams — and your chance to win some epic prizes.

Unwind with your peers for a couple of drinks on us!

See you at Enterprise AI Global East Coast in New York on October 14, 2027.

Beat the rush and join us early for complimentary barista-made coffee and breakfast.

Kick off the day with a welcome from your MC and a look at what's ahead across both streams.

As enterprise AI moves beyond experimentation, this session will examine how leading organisations are shifting from isolated use cases to enterprise-wide operating models that connect data, governance, automation and accountability to deliver measurable business value at scale.

• How AI leaders are moving from fragmented pilots to enterprise AI operating models that support scale, speed and measurable outcomes
• What organisations must align across data, risk, technology and business teams to embed trusted AI and automation into core workflows
• Where 2027 will separate leaders from laggards, from agent-enabled workflows and AI governance to workforce redesign and ROI accountability

This keynote will explore how evolving Australian standards, regulation and policy expectations are reshaping how enterprises design, govern and deploy AI with confidence.

• What the latest Australian policy direction means for enterprise AI governance, accountability and risk management
• How standards and compliance requirements are influencing data use, model oversight and vendor selection
• What leaders should do now to build AI systems that are auditable, responsible and ready for change

This panel will examine why so many AI initiatives stall after the pilot stage, and what enterprises need to get right in data, governance, operating model and talent to turn promising experiments into durable production capability.

• What changes when AI moves from sandbox to scale, from proof-of-concept enthusiasm to production-grade reliability, support and accountability
• How resource gaps differ between large enterprises and organisations early in their AI journey, and what each group can realistically prioritise next
• Where leaders should invest first when budgets, skills and internal support are uneven, so progress does not depend on a handful of AI champions

Hear how an enterprise reduced fear, confusion and resistance around AI by addressing the concerns of employees, executives and customers with clear communication, visible safeguards and hands-on support.

• How the team identified where hesitation was coming from, whether it was job security, trust, unclear value or fear of change
• What messages, education and engagement tactics helped executives, employees and customers feel more confident about AI adoption
• The change management lessons other organisations can use to make AI feel useful, safe and worth embracing

Recharge with refreshments and structured networking with your peers.

A hands-on, interactive session working through a real enterprise AI scenario as a room. Details announced soon.

Hear how one organisation tackled unclear AI ownership, inconsistent controls and rising compliance pressure to make enterprise AI safer and more scalable.

• How the team identified the governance gaps blocking AI from moving beyond pilots
• What practical controls, standards and workflows were introduced to keep AI deployment moving without slowing innovation
• The lessons learned for other organisations trying to build trust in AI while managing regulation, privacy and internal risk expectations

Learn how simplifying a crowded AI tool landscape reduces duplication and confusion, and creates clearer ways of working that made teams faster and more effective.

• How organisations can identify overlapping platforms, duplicated workflows and hidden waste across their AI and automation stack
• The practical steps needed to rationalise tools, streamline ownership and reduce confusion for business and technology teams
• What metrics and operating changes prove the simplification is working, from lower costs to better adoption and clearer accountability

This panel will explore how organisations can bring employees, executives and customers along the AI journey by aligning communication, trust, training and leadership at every stage of adoption.

• How to build executive confidence in AI programs so leaders champion change
• What employees need to hear, learn and experience for AI adoption to feel useful, safe and relevant to their day-to-day work
• How to manage customer expectations when AI changes service, decisioning or engagement models, without damaging trust
• The change management practices that help AI programs stick, from clear ownership and phased rollout to feedback loops and measurable adoption

Small-group, discussion-based sessions where you'll work through real enterprise AI challenges with peers in similar roles. Roundtable topics will be announced soon.

Enjoy a complimentary lunch while connecting with fellow attendees.

Put your knowledge to the test in this fast-paced quiz covering real-world trivia, key concepts, and emerging trends. Compete for bragging rights — and a voucher — as the top scorer takes the crown.

Explore how leaders can evaluate AI programs honestly, recognise when a use case is underperforming, and make the hard call to pause, simplify or scale down before more time and budget are lost.

• The early warning signs that an AI program is not delivering, from weak adoption and poor data quality to unclear value and rising operational drag
• How to build a practical review framework for deciding whether to fix, pause or stop a program, rather than endlessly funding a weak initiative
• How to reframe scaling down as a strategic decision that frees resources for higher-value AI use cases

This keynote will explore why education, understanding and training are essential to successful program adoption, and how organisations can equip both employees and customers to use new tools with confidence.

• Why training is not a tick-box activity, but a core enabler of adoption, value realisation and long-term program success
• How to design learning that meets different needs across workforce roles, leadership levels and customer segments
• What gets in the way of understanding new programs, from jargon and poor communication to lack of confidence and inconsistent support
• How organisations can turn education into a competitive advantage by building trust, improving usage and reducing resistance to change

This think tank will examine how organisations can align AI priorities, governance and execution across teams, and how to make responsibility for each program clear from the start.

• How to define ownership for AI programs so business, technology, data and risk leaders know who is accountable at each stage
• What governance model helps avoid duplication, competing priorities and shadow AI initiatives across the enterprise
• How leadership teams can set shared standards for data, security, compliance and value tracking without slowing delivery
• What practical decision rights should sit with central teams versus business units when AI programs move from idea to implementation

Wrap-up of the day's key takeaways across both streams — and your chance to win some epic prizes.

Unwind with your peers for a couple of drinks on us!

See you at Enterprise AI Global East Coast in New York on October 14, 2027.