Hello, Agent! A podcast on the agentic enterprise

The AI agent revolution isn't “coming,” it's already being built. In early 2025, only 5% of enterprise AI projects were in production. Now in 2026, 74% of executives report achieving ROI in their first year.

This is “the year AI gets to work,” and every leader in every industry is facing similar pressures and high expectations to get agentic systems into production. For enterprise AI, the stakes are even higher. It’s not just about building AI agents, but how to run them at scale safely and securely.

That’s why we created Hello, Agent!—the podcast that cuts through the hype and focuses on production-ready autonomous systems: how they actually work and how pioneers in enterprise AI are addressing universal challenges around data, infrastructure, and security in the real world. 

Hosted by Redpanda founder and CEO Alex Gallego, each episode features the AI leaders and builders shipping agents into mission-critical systems at scale (not just talking about it). These are the people defining what enterprise AI looks like—and we’re bringing you into the conversation.

Tune into episode one where we sit with Cyborg CEO, Nicolas Dupont, to discuss one of the biggest challenges: building secure architectures for enterprise AI agents.

‍

{{featured-podcast="/components"}}

‍

If you want a bit more context before diving into the episode, read on.

Behind the mic: meet Nicolas

Nicolas Dupont’s philosophy is that “privacy is not a privilege, but a fundamental right”—and his entire career is a testament to it. After founding his first cybersecurity company at 14 years old (as you do), Nicolas devoted the next 12 years to protecting data privacy and securing data infrastructure from the inside. 

With an initial focus on data compression before pivoting to data privacy and encrypted search, Nicolas built the foundation of Cyborg and its core product CyborgDB: the world’s first confidential vector database built for strictly regulated industries.

As a pioneer in the cybersecurity space, Nicolas has the rare combination of technical expertise and strategic savviness to navigate one of the most critical challenges in enterprise AI: how do you secure sensitive data when AI agents need access to it?

Sneak peek: securing vector databases for AI agents

For the unfamiliar, vector databases make it possible for machine learning models to quickly find similarities, identify relationships, and understand context. They store numerical representations of various types of data, like text or images, into an array of numbers (vectors). These representations are known as vector embeddings, which machines can then easily query for an exact match or similar data points. 

As Nicolas puts it, “Vector embeddings are the lingo franca of LLMs and transformer models today.” However, for many programmers, vector embeddings “seem safe” because they look like hashes—just a bunch of numbers. This intuition is dangerously wrong.

"Because you're encoding the meaning there, and you're able to do so relatively easily, it makes sense that you would be able to decode the meaning back out of those vector embeddings,” Nicolas explains.

In this first episode, Alex and Nicolas take a deep dive into the technical realities of vector embeddings, the shocking vulnerabilities in current vector database architectures, and the cryptographic innovations that make it possible to run Retrieval-Augmented Generation (RAG) agents on regulated data without creating chaos.

Note that his insights aren’t theoretical. They're drawn from real implementations with Fortune 500 companies—and now they’re all explained in his episode so you can use them, too.

Tune in and subscribe

This inaugural episode of Hello, Agent! sets the stage for a promising series on everything you need to know about deploying AI agents in production. From real-world examples to technical deep dives, this podcast is a must-listen for anyone in the “AI trenches.”

The future of enterprise AI is being written. Tune in to Hello Agent! to get ahead of it.

Coming up next: 

• Dominik Tornow (Resonate HQ): Simplifying agent autonomy at enterprise scale
• Jeremy Edberg (DBOS): Maintaining agent state when servers crash‍
• Tianshu Yu (ByteDance): Building AI infrastructure that supports millions of users

In the meantime, subscribe to get notified when new episodes drop.