Why Is Every Fortune 500 Company Sending Their CTO to the Same Hotel in May?
$1500 a ticket
There is a conference happening on May 4th and 5th at the New York Hilton Midtown that most people have not heard of. The AI Agent Conference.
Last year, 300 people showed up. This year, over 2,000 are expected. Every sponsorship package is sold out. The speaker list reads like someone merged the org charts of Thomson Reuters, GitHub, OpenAI, Bloomberg, Vanguard, Akamai, and Upwork into one room.
And yet this event barely registers on most people’s radar.
That gap between who is showing up and who is paying attention tells you something important. The people building the future of enterprise software have already decided what it looks like.
Read my previous letter: (Finding NemoClaw?) [2026]
The rest of the industry is still catching up.
If you build things, here is why this one matters…
THIS IS NOT ANOTHER “AI” CONFERENCE
There are hundreds of AI conferences every year. Most of them cover everything: foundation models, ethics panels, research papers, robotics demos, and somewhere in the back of the schedule, a session or two about agents.
The AI Agent Conference is different because it is the only major event where agents are the entire agenda. Not a track. Not a side stage. The whole thing. Three tracks across two days, all focused on one question: what happens when AI stops answering questions and starts doing work for corporations at scale?
The three tracks are Agentic Enterprises (how autonomous AI changes business operations), Agentic Engineering (how you actually build the infrastructure), and Agentic Industries (what this looks like in finance, healthcare, legal, and logistics specifically).
That structure alone tells you where the market has moved. A year ago, “AI agents” was a buzzword people dropped into pitch decks. Now there are enough real deployments, real failures, and real lessons learned to fill 2,000 seats and three parallel tracks for two days straight.
THE SPEAKER LIST IS THE SIGNAL
Conference lineups are usually a mix of big names smelling their own farts — level keynotes and practitioners buried in afternoon breakout sessions. This one is unusual because the big names are the practitioners.
Steve Hasker is the CEO of Thomson Reuters. Not a VP of innovation. The CEO. Thomson Reuters processes legal and financial information at a scale that touches nearly every major institution on the planet. When their CEO shows up at an agent conference, it is because autonomous AI is not a research project inside his company. It is the strategy.
Robert Blumofe is the CTO of Akamai, the company that handles a significant percentage of global internet traffic. Mario Rodriguez is the Chief Product Officer at GitHub, which means he is the person deciding how AI agents get integrated into the tools that developers use every single day. Andrew Rabinovich is the CTO and Head of AI at Upwork, a company whose entire business model gets reshaped by agents that can do the work that freelancers currently do.
Then there is the builder side. Joao Moura, the co-founder of CrewAI, one of the most widely adopted open-source agent frameworks. Arvind Jain, the founder of Glean, which just became one of the fastest-growing enterprise AI companies in history. Chris Kyle from OpenAI. Tomas Reimers, the CPO of Graphite, which was acquired by Cursor. Nick Schrock from Dagster Labs, who also created GraphQL.
And representatives from Datadog, Google, Meta, Microsoft, Citigroup, AWS, and Databricks filling out the rest.
The reason this matters is not name recognition. It is that the people in that room represent both sides of the same equation. The executives who need agents to work at enterprise scale and the engineers who are building the systems to make that happen. Most conferences have one or the other. This one has both, by design.
Simon Chan, the managing partner of FirsthandVC who curates the event, put it simply: “The biggest breakthroughs come when corporate leaders and the technologists building AI actually sit down together.” I tend to agree.
FirsthandVC is not a conference company. It is an inception-stage venture fund whose entire portfolio is built around autonomous agentic AI. CrewAI, Distyl AI, Eudia, and a dozen others. The conference is an extension of an investment thesis, not a media play. That changes the quality of what gets discussed on stage.
THE AGENTIC LIST TELLS YOU WHERE THE MONEY IS GOING
Before the conference, the organizers partnered with NYSE Wired to release something called The Agentic List 2026. It is a curated ranking of the top 120 private companies building enterprise-grade agentic AI. Here is the link: [Agentic List 2026]
The numbers behind it are worth understanding. Over 5,000 nominations came in from industry executives and investors. Nearly 2,000 companies were screened. 120 made the cut. The evaluation weighted real production deployment and executive validation more heavily than funding or hype. Their words: “Industry adoption and executive validation carry the heaviest weight, reflecting real production impact over hype.” Big words.
The 120 companies are organized across 14 categories spanning the same three pillars as the conference tracks. Agent development platforms. Agentic infrastructure and data systems. Safety, alignment, and observability. Coding and engineering productivity. Vertical applications in finance, healthcare, law, retail, and logistics.
Some recognizable names made the list. Ramp at $2.8 billion raised. Mistral AI at $3.2 billion. Cohere at $1.5 billion. Perplexity at nearly a billion. But the list also surfaces earlier-stage companies that most people have not heard of yet, companies operating in categories like “Safety, Alignment, and Observability” that barely existed as a market two years ago and are now considered critical infrastructure.
If you are trying to understand where capital is flowing in the agent space, this list is one of the clearest maps available. Not because it is definitive, but because it reflects what 5,000 executives and investors actually nominated when asked “who is building the real thing.”
WHAT BUILDERS SHOULD ACTUALLY TAKE FROM THIS
Here is the part that matters if you write code, ship products, or make technical decisions for a living.
The shift from demo to deployment is happening faster than the discourse suggests. The AI conversation online is still dominated by model benchmarks, viral demos, and philosophical debates. Meanwhile, Thomson Reuters, Vanguard, Akamai, and Citigroup are sending their most senior technical leaders to a two-day conference about how to deploy agents in production. These are not companies that attend events for inspiration. They attend because they have problems to solve and need answers now.
The infrastructure layer is where the opportunity is. The conference has an entire track called Agentic Engineering. Not “how to prompt an agent” or “look what this agent can do.” Engineering. Infrastructure. The unsexy stuff that makes agents actually work at scale.
If you are a developer looking at the agent space and wondering where to focus, the existence of that track is your answer. The world does not need more agent demos. It needs better agent infrastructure: (NemoClaw is my pick atm) orchestration, observability, failure handling, security, and state management.
Vertical-specific agents are the next wave. The Agentic Industries track covers finance, healthcare, legal, and logistics as separate domains with separate problems.
This is important because it means the market has moved past “general purpose agent” as a product category. The companies getting traction are the ones building agents that understand the specific workflows, compliance requirements, and data structures of a particular industry.
A legal agent and a healthcare agent are fundamentally different products even if they use the same underlying models. The companies on The Agentic List that raised the most money are almost all vertical-specific.
YOU DO NOT NEED TO BE IN THAT ROOM TO ACT ON THIS
One more thing worth saying.
A $1,500 ticket to a conference in midtown Manhattan is not accessible to everyone. But the underlying shift it represents is.
The agent frameworks being discussed, CrewAI, LangGraph, Autogen, and others, are open source. The models that power them are increasingly available as open weights. Google released Gemma 4 this week, a model family purpose-built for agentic workflows that runs on consumer hardware.
An RTX 5090 with 32 gigabytes of memory can run quantized 70-billion-parameter models locally at speeds that make real agent workflows practical.
The infrastructure categories on The Agentic List, agent development platforms, observability tools, safety and alignment systems, those are all areas where individual developers and small teams can build meaningful products. You do not need enterprise sales to build agent tooling. You need to understand the problems that enterprises are hitting, problems that are being discussed openly at events like this one, and build solutions.
The conference is a window into what the largest companies in the world are struggling with right now. That struggle is your opportunity.
What do you think?
God-Willing, See You at the Next Letter
GRACE & PEACE