Thank you for joining us last week at Aria for what turned out to be an exceptionally candid conversation with old friends and new about the AI adoption challenges we're all navigating.
The quality of discussion and debate around the table, including on what it’d take to actually realise the vision of a 3 day work week, reinforced why you’re all leaders in the Australian AI landscape.
It was really valuable hearing some of the war stories of what it’s taken to scale AI in your respective organisations, with common themes emerging from banking and technology, to manufacturing, aviation and even the government sector. It’s clear that we’re at an inflection point in AI adoption and investment in the enterprise market, and we’re excited for what 2026 has in store.
We touched on several recent client solutions and examples of how brands globally are focusing their AI investment in strategic areas, which are all unique to their business strategies. We wanted to share a few more details with the below examples, and would love to deep dive into any other particular areas of focus or challenges you’re seeing.
We’re looking forward to connecting with each of you in the coming weeks, and on behalf of the entire OpenAI & Tomoro teams we hope you all enjoy a well deserved Christmas break with loved ones.
Brand aligned, personal travel concierge with speech-to-speech for Virgin Atlantic
Virgin Atlantic customers can plan, book & experience travel differently through an AI concierge built to harness the company’s knowledge and deliver the airline’s signature warm tone to more people.
The concierge intelligently combines speech, text and visual LLM capabilities to deliver a tailored experience for every customer. Traditional deterministic booking systems fail to adapt to different lines of questions. We’ve designed and built a Concierge experience that actually listens to guests, responds in a way that feels natural and curates and visualises travel plans in a way you simply can’t do traditionally.
Bespoke, multi-modal customer support for Supercell
In just 11 weeks, we built and deployed a multi-modal, virtual agent solution that classifies messages, flagging any particular risks or complex cases for the specialist human agents to handle, and generates a response tailored to a player's specific context.
And when support tickets peak along with interest in games - sometimes at 40,000 tickets a day - the virtual agents can handle the vast majority autonomously. With an average response time of 7 seconds -down from up to a day - and all within the gameplay experience and to the same level of accuracy, players are delighted. CSAT scores have jumped by 20%.
With a 90% cost reduction in handling tickets, the burden on Supercell’s highly trained human support team is reduced and those who really understand the game can focus on producing better outcomes on the most complicated cases.
Turning Supercell's in-game store support into instant, on-brand conversations
Now the first point of contact for any in-game purchases, Supercell’s Store Bot resolves ~70% of tickets instantly, from missing purchases to refunds, while escalating complex cases to specialists.
The multi-agent system classifies enquiries, retrieves the right info, and a dynamic persona adjusts tone to fit the player. It factors in live store offers and player context like location to deliver tailored recommendations in human-like interactions. Guardrails keep responses safe across payments, legal, and data queries.
The solution reduces agent workload, brings faster response times and delivers a consistent, on-brand player experience across all stores.
Real-time insights agent for live NBA commentary at scale
Our solution augments real-time NBA data with LLM-generated commentary, to deliver tailored highlights to users in a basketball specific tone-of-voice. The multi-agent system surfaces highlights, ranks events and provides commentary within 5 seconds.
Users receive highlights based on their specific player and team interests, with 25,000 insights delivered to 2 million users.
In the future, users can ask the agentic questions on the complex data, with the system retrieves information and answers these queries.
