The biggest takeaways from ScaleUp:AI 2025: Agents, automation, and the changing roles of humans

With another ScaleUp:AI in the books, we close another day of thoughtful conversations and productive connections. This event invites founders, builders, enterprise execs, investors, and researchers to gather for thought-provoking discussions, and this year, we had 3,000+ registrants from over 70 countries and welcomed nearly 400 in-person attendees.

Below, find some key themes and quotes from the day. Registrants, you’ll get the recorded sessions delivered to your inbox soon. If you managed to miss us entirely — be sure to subscribe to the Insight Partners YouTube channel when we publish select sessions to the public later this year. 

The agentic enterprise is taking shape

In his opening keynote, Jeff Horing, Insight Partners’ cofounder and managing director, framed the moment succinctly:

“The difference between software and service is going to look a lot more blurred.”

AI is no longer an add-on — it’s becoming the connective tissue across workflows, products, and customer interactions. Panelists, including Itai Asseo (Salesforce), Joao Moura (Crew AI), and Ritika Gunnar (IBM), expanded on how agentic AI is reshaping the enterprise stack. Asseo predicted that “ambient intelligence is going to be part of every process we do,” while Moura noted that leading companies “are thinking of AI agents as infrastructure.”

Process becomes the new moat

Ethan Mollick (Wharton) reminded attendees that AI advantage will come from execution, not model size.

“Moats come from some sort of process that makes a difference.”

In conversation with Lonne Jaffe (Insight Partners), Mollick explored how threshold effects in generative AI are shifting competitive dynamics. Companies that rewire workflows and scale human-AI collaboration — what Horing called the “80% world” — will outpace those chasing theoretical perfection.

Humans still anchor the loop

Across sessions, leaders emphasized that human judgment, accountability, and creativity remain irreplaceable. From Lt. Gen. Jack Shanahan, Inaugural Director of the Department of Defense’s Joint AI Center:

“Human control is a given. The question is how, when, and where that human control will be applied.”

Esther Dyson and Ethan Mollick echoed that sentiment from different angles: prompt engineering as a “people skill,” and the need for human oversight in agentic systems. Even as AI autonomy accelerates, trust and accountability are the scaffolding that make innovation sustainable.

Reliability, security, and ethics as the new infrastructure

From May Habib (Writer) to Alexis Bjorlin (NVIDIA), speakers called for a deeper focus on reliability and security in the next generation of AI systems. Habib stressed that “it’s especially important to know what is highly reliable when everything now looks the same.”

On the infrastructure front, Bjorlin drew a sharp analogy.

“We’re moving from the traditional manufacturing of goods to the manufacturing of intelligence.”

Panelists agreed that agentic systems demand “full-stack” security and new forms of governance — a foundation that balances innovation with resilience.

Tony Fadell, founder of Build Collective and the design team behind the iPod, iPhone, and Nest, reminded the audience that innovation means little without trust and usability.

“Before my daughter could read, she interacted with Siri,” he said. “We’ve taken out all the friction… so anything that is somewhat interesting and basically free can get adopted everywhere.”

Fadell urged founders and investors to remember that every AI interface — no matter how intelligent — still meets a human at the other end.

Transformation is costly — but delay is costlier

Marc Boroditsky (Nebius) warned that “if enterprises don’t transform early enough, they’ll transform under threat.” Rob Ferguson (Microsoft) underscored that “specificity is the key to success” when deploying AI projects — vague ambitions won’t survive real-world complexity.

Several speakers, including Lynn Martin (NYSE) and Esther Dyson, pushed for long-term, principled approaches to corporate governance and technology adoption, arguing that sustainable transformation requires both urgency and foresight.

Building the next AI capital of the world in NYC

This year, Insight is pleased to support the AI innovation happening in our hometown of NYC. The event spotlighted AI’s role as an economic catalyst. Andrew Kimball (NYCEDC) announced pilot programs in AI investment and literacy at ScaleUp:AI with the announcement of NYC AI Nexus operators:

“NYC is cementing our position as the AI capital of the world.”

Read the press release and learn more: NYCEDC Announces Operators of NYC AI Nexus to Advance Applied AI Venture Creation and Equitable AI Adoption Across New York City’s Economy.

Thank you to our partners and speakers

Thank you to our 32 exec-level speakers and experts and 31 leading AI partners for making this event possible:
Adobe, Dell, Google Cloud, iCapital, Morgan Stanley, NYSE, Oracle, Rackspace, SAP, SVB, Willkie, Acceldata, a.team, Copado, CrewAI, Exiger, HoneyHive, Incident.io, Kiteworks, Lightrun, Marketing Evolution, Postman, Precisely, Promptfoo, Runware, Shelf, Skyflow, Valence, Writer, Riviera, T200.

You can find special offers, thought leadership, and special offers at the ScaleUp:AI website.