We just wrapped Samba Summit 2026. Three days just outside of San Francisco with our largest partners and clients, including platforms like Google, Meta, Netflix, OpenAI, Amazon, and TikTok, alongside some of the most brilliant marketers, technologists, and storytellers in Hollywood.
Art vs. Science: The Debate of Our Generation
Some people see media as an art. It’s all creative judgment. Others see it as a science, a math problem waiting to be solved. The industry has spent twenty years letting those two camps fight, and the fight has not made anyone’s work better.
That tension was the spine of this year’s program, amplified by a news cycle dominated by AI and the risk it places on white-collar work, industries, and civilization itself.
We embraced that tension head on with a chat with Alan Taylor, whose work includes Game of Thrones, The Sopranos, Mad Men, and Deadwood. We took his reactions to the craze around AI-generated microdramas in a vertical format on your phone.
We had Adam Cheyer, founder of Siri, make the case for slowing down AI innovation so we can figure out what the end goal and purpose is for all of us.
Samba put its mission and vision on display for our partners. Our mission: support art with science, and enable creators to thrive across all platforms — engaging their audience everywhere.
I remain an AI and technology optimist, with one condition I’ll come back to at the end.
Why Did We Start Samba?
Before Samba, I would come home, turn on the TV, and scroll through hundreds of channels without finding anything that moved me. The stories were out there. The distribution model was the problem. We bet that internet-connected television would fix it, and it did.
But it created a new problem for everyone in that room. For seventy years, media was a handful of heavily monetized networks reaching enormous audiences at once. Now the content and the audiences are scattered across more platforms and formats than any human planner can hold in their head. No brand and no creator can reach their full potential on one platform anymore. The best ones invest everywhere: television, streaming, social, retail media, and whatever comes next.
Activating all of that in a coherent way is the job now. We call it media intelligence. Understanding the content, understanding how audiences actually engage with it, and understanding what brands are doing about it. Those three raw materials, connected, are what the Samba Knowledge Graph is built from: more than 400 billion consumption signals a month across 1.5 billion people.
The takeaway from the room was not that anyone disputes this. It’s that almost nobody has the plumbing to act on it.
Every Marketer Becomes a 10x Marketer
We ran two hands-on breakout tracks this year. One had attendees building audiences with an agent. The other put them in front of our data with an AI harness and asked them to build analytics dashboards. No code. No SQL. Just a conversation with the data.
Our industry has long celebrated the 10x engineer. The rare person who could ship in a week what a team needed a quarter to build. Companies were organized around finding them, keeping them, and staying out of their way.
With AI coding agents, every engineer became a 10x engineer. The ceiling that used to belong to a handful of people is now the floor.
That is exactly what we saw in the breakout. People who had never written a line of code or a single query built real dashboards on real data in under thirty minutes. The constraint was never talent or curiosity. It was the tooling.
With Samba’s tools, we believe every marketer will become a 10x marketer.
Think about what a marketer’s week actually looks like today. Waiting on an analyst to pull a report. Waiting on an agency to turn a brief into a plan. Waiting on a vendor to confirm an audience was trafficked. Stitching together numbers from five dashboards that each count the same household differently. Most of the job is waiting and reconciling, and the small window that’s left is where the actual marketing happens.
Now imagine that window is the whole week. You ask a question of the data and get an answer in seconds, not days. You describe an audience and the segments build themselves. You set an objective and a plan assembles across every platform, deduplicated and ready to go. The agents do the waiting and the reconciling. The marketer does the thinking.
That is what 10x means here. Not ten times more output from the same people, but the same people spending ten times more of their energy on judgment, creativity, and strategy, and ten times less on the plumbing. The engineers got there first because the tools arrived for them first. The tools for marketers are arriving now, and we intend to build the best of them.
Samba Has Emerged as an Agentic Advertising Company
Attendees of the Summit got to see Maestro — our orchestration agent that works with marketing teams to help them identify their audiences and execute on their marketing objectives. Maestro has subagents for every step of the media planning lifecycle.
Our clients at the Summit expressed desire to use these agents at every step of the marketing funnel. They want to move past the old division between reach and performance. They want both objectives executed at once, in the same plan, and we demonstrated how agents can analyze and optimize against both together rather than trading one off for the other. They also wanted Samba’s identity infrastructure (Project Gravity) to help de-duplicate all of the media vendors in the plan, to ensure that every new impression is incremental to the last and that every audience is reached at the right level of frequency.
First Party Data Is Your Advantage
Marketers are rightly nervous about handing data and decisions to software. The thing that earned the most nods in the demo was not the creative generation or the plan builder. It was the connectors to Databricks, AWS, and Snowflake.
Here is why that matters. Every brand in the room is sitting on first-party data they have spent years and real money assembling: customers, transactions, loyalty members, site visitors, CRM records. It is the most valuable asset they own and the one they are least willing to move. The moment a vendor asks for a copy of it, the conversation slows down, legal gets involved, and most of the value never leaves the building.
The connectors flip that. The data stays where it lives, inside the client’s own Databricks, AWS, or Snowflake environment, under their security and privacy controls. The agents reach in with permission, join the client’s first-party data to the Samba Knowledge Graph, and carry the result straight into marketing activation and analytics. No copies. No exports. No new place for the data to leak from.
That is what makes first-party data an advantage rather than a liability. When your customer records can be matched to actual viewing and browsing behavior across every screen, and that match can be activated on every major platform and measured independently, the data you already own becomes the most powerful input in the plan. Agents that cannot reach it are planning blind. Agents that can are planning with everything you know.
The other thing that builds trust is anchoring all of your AI in truth.
A language model on its own is a very confident guesser. Point it at a pile of raw tables and it will give you an answer, and the answer will sound right whether or not it is. That is fine for a brainstorm. It is not fine for a media plan with real money behind it.
The fix is an ontology. A formal map of your world that says what a household is, what a show is, what an ad exposure is, what a conversion is, and how each of those relates to the others. The ontology organizes the data, gives every piece of it a label that a model can understand semantically, and turns a warehouse full of rows into knowledge a system can reason over. That is what the Samba Knowledge Graph is underneath: our data, organized by an ontology, so that when an agent asks a question it gets a grounded answer instead of a plausible one.
That is also what makes AI reliable enough for enterprise applications. The agent is not inventing a reach number. It is reading one, from a graph that knows which households saw which ads on which screens. When the plan says 24 million deduplicated households, you can trace that figure back to the signals that produced it.
That is how you let agents work on a brand mission with high stakes. Every action, every number, every decision ties back to ground truth and real data. The bar is simple: a marketer should be able to defend every decision the system made to their own CEO or CFO.
Our Call to Action: Harness AI, Don’t Serve It
The tech marketing industry has been using scary positioning statements about technology since the 1999 launch of the Power Mac G4, which Apple advertised as so powerful the government didn’t want you to have it.
The AI industry has picked up that playbook in recent months to differentiate one model from the next. It has gone further, describing the devastating effect its products will have on every other job and industry. Wall Street loves that narrative. It justifies massive multibillion dollar investments to support it.
I believe it’s given every marketer, storyteller, and creator an imperative. We must lean into the mission, values, and creativity that define us.
Inspire people to do the best work of their lives, and harness the power of AI rather than become a slave to it. The world needs inspiration, clarity, and creativity now more than ever.
Whether you’re a creator, a media company, or a brand, there is no replacement for human wisdom and judgment. Technology exists as a set of tools to amplify a message that should improve the human experience.
Now is the moment when the creative industry should be working hardest and fastest to demonstrate not only how to coexist with AI, but how to harness it as a force for good.
We started this company because we believed great creative work deserved a better chance of finding its audience. For the first time, we have the technology to do that end to end. The people in the room this week told us exactly where it falls short. That is the most useful thing a summit can do.
One last thought, which I shared from the stage.
Humans have a superpower: our creative spark, and the judgment and discrimination that let us separate the signal from the noise. We have to use it to cut through the AI slop.
AI is, at best, a historian. It can see everything that has already happened. A historian can be very helpful, because if there is a flaw in human nature, it’s that we repeat our past mistakes. The best use of AI is to prevent that from happening. The best use of our hearts and minds is to guide us forward.
May we use our human gifts and talents to lead us into uncharted territory, always staying one step ahead of the technology, and remaining the masters of our own destiny.
Thank you to everyone who came. See you next year.