Most AI assistants still start every conversation as strangers. They forget the customer they spoke to yesterday, the decision the team made last week and the preference you have explained three times. That is changing quickly. A new generation of agents can carry memory across tasks, learn how a team works and pick up a project exactly where it was left.

This keynote explores what memory makes possible and what it demands. How remembering turns a clever tool into a dependable colleague, which memories an enterprise should keep, share or deliberately forget, and how leaders protect privacy and trust when their AI never truly closes the file.

Here is a simple test for any AI program. If your main model provider doubled its price, changed its terms or became off limits in one of your markets, how long would it take to move? For many organizations the honest answer is months, because prompts, evaluations, fine tuning and workflows were all built around one vendor without anyone deciding that on purpose.

This keynote panel treats portability as a discipline rather than a hope. Our panelists share how they run switching drills, which parts of the stack they insist on owning, the contract clauses that protect their exit, and how they balance the convenience of one provider against the freedom to leave.

The race for ever larger models makes the headlines, but inside many New York banks, insurers and asset managers a quieter story is unfolding. Smaller models trained for a single domain are proving faster, cheaper and easier to govern, especially for numerical work like forecasting, pricing and risk, where a confident guess is worse than no answer at all and a model risk team will want to see every step.

Our panelists share where they swapped the giant for the specialist, what it did to accuracy and cost, and how to decide which problems truly need the biggest brain in the room.

New York has already put AI rules in place that touch everyday enterprise work. City employers must commission independent bias audits for AI tools used in hiring and promotion, and enforcement is tightening after a critical state review of the program. Insurers are expected to show that AI used in underwriting and pricing is fair. And from January 2027, the developers of the most powerful AI models face new safety and transparency duties under the RAISE Act.

This panel helps leaders see how these rules fit together, what they ask of companies that buy AI rather than build it, and how to turn New York's early requirements into a governance model that will travel well as other states and countries follow.

The phone call never died. It is still the channel customers reach for when something truly matters, and for years automation there meant endless menus and the familiar plea to speak to a human. A new generation of voice AI reasons directly on speech, following context, tone and intent across an entire conversation.

This session explores what that means for customer service, sales and internal support. Where it already works, how to design it with dignity, and how to keep trust when the voice on the line is not a person.

Some of the most valuable data in the enterprise is the data you are not allowed to touch. Patient records, customer conversations, fraud cases, the rare failure that happens once a decade. Synthetic data offers a way through, realistic records that carry the patterns without the people, ready for testing, training and sharing.

This fireside chat separates the promise from the pitfalls. Where synthetic data is already speeding up projects in regulated industries, how to prove it is both useful and truly private, and what happens to models that learn too much from machine made examples and too little from the real world.

Your agents are starting to call tools, negotiate with other systems and hand tasks to agents built by partners and vendors. Behind every one of those conversations sits a new layer of protocols and connectors that most enterprise architectures never anticipated.

This session demystifies the emerging standards that let agents work together, and shows how to adopt them with clear rules for trust, logging and accountability from the very first day.

Most of the AI conversation has lived on screens, in chat windows, documents and dashboards. The next chapter steps into hospital corridors, building systems, warehouses and the last mile of delivery through some of the busiest streets in the country. World models help machines understand space, time, cause and effect, so robots and autonomous systems can rehearse an action before they take it.

Our panel explores where physical AI is already creating value in a city as dense as New York, what it demands from data and infrastructure, and how leaders prepare their operations, safety teams and people for machines that move.

Every organization has a few people who simply know. The banker who remembers how the last crisis really unfolded, the underwriter who senses a bad risk before the numbers show it, the operations lead who knows why a rule was written twenty years ago. Much of what they know was never written down, and many of them are approaching retirement.

This closing conversation explores how leaders are using AI to listen to, record and preserve that expertise while those experts are still in the building. How to invite people to share what they know rather than feel replaced by it, how to turn stories and judgment into something a new colleague or an agent can learn from, and how to honor the people whose wisdom built the company.