What is actually shipping in Generative AI in 2026? Eric will set the stage for half a day of talks, panels and hands-on sessions on applied Generative AI at the iconic Curzon Soho cinema – from how the technology is reshaping marketing and transforming creative industries, to the latest wave of agentic AI and the opportunities and challenges of deploying enterprise GenAI at scale.

For twenty years, brands have lived safely on their island — fixed fonts, locked hex codes, a defended tone of voice. Exploration was expensive and risky, so we optimised for safety. But AI has collapsed the cost of creation, and that's exposed a trap: if every brand uses the same advanced tools to optimise for the average, we all drown in the same sea of sameness. The real risk is no longer stretching too far and looking silly — it's never leaving the island at all.

This talk introduces Real-Time Creative, a way of working where the tools of imagination and the tools of production have merged into one shared space — strategists, copywriters, filmmakers and AI artists prototyping what a brand could be across a thousand contexts, and testing it in days, not quarters. You'll see it in real campaigns, from a skeleton "Grimfluencer" invented for a non-alcoholic beer to a luxury fashion campaign extended on the fly and a space-tourism film shot without leaving the studio. But the payoff isn't the tools — everyone gets those. It's the structural move that decides whether a brand explores or stagnates: who you send out to do this, and how you give them permission to get a little lost. Let's build the raft.

By 2030, the vast majority of content will likely be AI-generated. AI-generated content is growing at around 40% per year, compared to just a few percent for traditional content. The overall content market will continue to expand, with AI-generated content set to overtake non-AI content. The legal landscape around human likeness is also evolving rapidly. As AI becomes essential to modern marketing, is your organisation ready?

The Neural Forest digital art project introduces a pioneering "phygital" approach that materialises digital concepts into tangible, premium video sculptures. While advanced generative AI tools expand the limits of imagination, their visual art outputs typically remain trapped as fleeting pixels on a screen. This project bridges that virtual-physical divide, transforming fluid digital files into high-value, physical art assets.

The production pipeline seamlessly moves from initial concept to digital fabrication. In a collaboration between human and machine, 3D mesh modelling is used, continuing through industrial 3D printing, and high-gloss chrome plating. And raw organic structures are transformed into reflective chrome-finished sculptures. These physical forms securely cradle LCD screens running custom video loop artworks made from a unique combination of photography, AI, and datamoshing.

Looking ahead, Neural Forest heralds the transition towards a metamorphic engine that bridges digital concepts and tangible reality. It points to a future of pure creation, where machines will materialise ideas into diverse physical forms, ranging from synthetic compounds, and eventually, the human form itself. This presentation demonstrates a sophisticated pipeline where AI and nature converge, offering an exclusive look at the future of high-end, tangible digital art.

As AI agents move from chat interfaces into creative tools, infrastructure, and decision systems, they are usually framed through productivity: better prompts, faster outputs, more autonomous assistants. This talk proposes a different view: AI agents as an expressive medium.

Drawing on my ongoing project Assembly, I will show how agentic workflows can generate artistic material by operating semi-autonomously — debating, interpreting, telling stories, evolving, and drifting through feedback loops. Rather than treating the final output as the artwork, the work focuses on the process itself: how meaning shifts, how narrative develops, and how the system accumulates distortions as it runs.

The talk also considers agentic systems as speculative laboratories for exploring wider cultural and cognitive processes — how they drift, break down, or produce distortions in real time.

The key takeaway is that AI agents in creative fields are not only tools for making images, texts, or videos. They can become instruments for staging, watching, and thinking through how meaning is produced, reinforced, distorted, and believed.

Gabriela will present an extract from her award-winning AI film 'Organising Principles of Experience' and discuss the process of developing a film in collaboration with an AI model tuned to the writings and sensibilities of experimental filmmaker Maya Deren (1917-1961).

You've seen it in your own feed: more content than ever, more polish than ever – and somehow less that stays with you. Anyone can generate now. The harder question is what makes something worth keeping.

The three artists on this panel have spent years inside that question – not theorising about AI creativity, but making with it every day. Guli Silberstein turns his AI films into physical sculpture, because some work deserves weight you can stand in front of. Vadim Epstein lets autonomous agents carry his work somewhere he didn't plan, and finds the art in watching where it goes. Gabriela Tropia made a film together with the voice of a director who died sixty years ago – and discovered something surprisingly human in that collaboration.

In conversation with Denis Samuylov, they'll share what working alongside machines has taught them about taste, authorship and the parts of creativity no model has touched yet. If you build with AI – or you've started wondering what your role is now that it creates too – this conversation is for you.

Most product managers using AI today are getting a modest efficiency gain, maybe 20%, by pointing it at repeatable tasks.

​That is useful, but it is not the step change everyone keeps promising.

​The PMs who learn to build a product brain and orchestrate agents, rather than just prompt them, will work at a completely different level: automating the repeatable work, and reserving their own time for the decisions that actually need human judgement.

​Emma Burrows is the CEO of Rezonant, an AI platform that helps product teams define and ship the right changes at speed. She was previously UK CTO at Stripe, where she scaled the London engineering team from zero to fifty, and has held leadership roles at Google and Charlotte Tilbury.

​In this session, Emma will cover:

• ​What "agent orchestration" actually means in practice

• ​Building a "product brain": the right inputs, automations and workflows

• ​What end-to-end orchestration looks like: where to automate, and where to add friction so a human stays in the loop

​This is a presentation followed by audience Q&A.

What happens when your AI agent encounters unexpected, adversarial, or incomplete inputs?

In this session, we explore how to proactively check your own AI systems to make them more robust. We will cover practical red teaming techniques including adversarial prompting, context manipulation, and tool exploitation. Using a demo in Microsoft Foundry, we will see edge cases on an AI agent. This talk is designed for practitioners looking to move from away from experimentation into production.

Most AI pilots impress in demos but fail when they meet the messy reality of teams, tools, permissions, workflows, and organisational context. In this 10-minute talk, I’ll share what we’re building at ollo: an enterprise context layer for agentic AI that helps companies move from generic assistants to agents that understand how work actually happens.

The core problem is not model capability alone; it is whether AI can operate with the right structural, semantic, operational, and live context while giving teams the confidence and capability to adopt it. I’ll show how ollo approaches this through context-aware agents and co-build sprints with non-technical teams, rather than one-off AI rollouts that never scale.

The key takeaway: the next wave of agentic AI startups will not just build more powerful agents. They will build the infrastructure and adoption models that make agents useful, trusted, and embedded in day-to-day work. Attendees will leave with a practical framework for thinking about enterprise agent deployment, and why context is the missing layer between AI prototypes and real operational impact.

AI agents are moving from demos into production and as they do, they stop being tooling story and start doing work end-to-end. Who deploys agents first, what work they take over, where the value lands, and how organisations transform while machines do more of the doing: that is the agent economy.

This session is a panel with Nilson Kufus (Co-Founder & CEO, Rebels) on agents completing real work end-to-end. He is joined by Akshat Goenka (VC, Moonfire) and Jounghyun Shim (VC, Breega): what separates durable agentic companies from impressive demos, where defensibility forms as model capability commoditises, where business models move to, and what investors look for between pre-seed and Series A.

The session also announces the London chapter launch of the Agent Economy Association: A cross-industry community connecting operators, builders, investors, and beyond, around the emerging agent economy, now expanding from Zurich to London.

Many organisations have deployed AI at speed, but adoption alone rarely leads to meaningful transformation. The missing piece isn't more tools — it's moving from solo AI to multiplayer AI: systems where intelligence is shared, context compounds across teams, and the whole organisation gets smarter over time. This session explores what it takes to go beyond individual productivity gains and build shared AI intelligence that delivers measurable impact at scale.

What organisations get wrong when measuring AI impact — and why tracking activity instead of outcomes masks the real problem

Why individual AI usage stays siloed — and how to design systems that learn collectively across teams.

The shift from solo to multiplayer AI: building shared agents with institutional memory that compound knowledge over time.

What "smarter over time" looks like in real workflows — and how leaders can architect for it.

Every financial institution wants to scale with AI. More are discovering that the constraint is not the technology but how these systems can be governed safely.

Boards and executive teams are approving AI systems, expanding use cases, and pushing for faster deployment. But as scale increases, so does the gap between what governance frameworks say and what leaders can actually do when something goes wrong. Decision rights were designed for human-paced execution, escalation pathways assume time to think, and intervention capacity that looks solid on paper has often never been tested under real pressure.

This is how the AI scale trap forms - and it forms faster than most boards realise. The firms caught in it are not those moving too slowly. They are those moving fast on governance foundations that were never built for machine-speed execution.

The firms scaling AI with confidence in regulated environments share a different approach. Authority is clear, escalation works, and named executives can demonstrate they could detect a problem and intervene before impact. Governance does not constrain - it is their competitive advantage, and the foundation for growth that is both consistent and safe.

This session makes the case that the path to sustainable AI scale runs through governance architecture. Drawing on two decades of senior leadership in global financial services, including regulator engagement and board-level accountability, it offers a framework for understanding why governance is where scale breaks down, what AI-ready institutions do differently, and how to turn oversight into competitive advantage - because the institutions that get this right are those where human judgment and structural control work together, and where the person accountable can act.

Generative AI has moved far beyond experimentation. Across industries, organisations are now deploying GenAI solutions into core business processes, products, and customer experiences. This panel brings together leaders at the forefront of enterprise adoption to discuss what it really takes to scale GenAI successfully – from technology, infrastructure, and architecture to governance, security, change management, and organisational readiness. Join us for an honest conversation about lessons learned, common pitfalls, and where enterprise GenAI is headed next.