Redpanda Data

1.1 trillion messages a day. 5.5x lower latency. Featured in Jensen Huang's keynote. Just another day for Redpanda. 🤘 NVIDIA Vera + Redpanda is the data layer enterprise agentic AI has been waiting for. NYSE and Hewlett Packard Enterprise already know. Now you do, too. Read the blog: https://lnkd.in/e_vuwz4a #NVIDIAGTC

💬 Hello, Agent: How do you make AI agents fail-safe? Learn the answer in Episode 3 of our "Hello, Agent!" podcast with the super savvy Jeremy Edberg, C-Suite Advisor at DBOS, Inc. In this ep, Jeremy explains how durable execution solves reliability issues, its role in building production-ready AI agents, and shares insights from his time at Netflix and Reddit on enterprise-scale reliability engineering. Tune in 🎧 https://lnkd.in/d8p4FNxt

LFG! Having the worlds best company NVIDIA talk about Redpanda Data and NYSE makes me infinitely proud to partner with everyone at Redpanda. Live now.

We first started by creating a duality of streams and [iceberg] tables. We integrated iceberg topics w/ all the main catalogs with zero ETL. We then added redpanda connect CDC as a way to combine operational data from oracle, postgres, mysql, microsoft sql, so you could freely query them in your lakehouse. TODAY you can spin up `psql` and rule them all. Huge congratulations to the Redpanda SQL team: https://lnkd.in/d2WeNSwA What took a long time is ensuring we had the right architecture to continue to change the permissioning systems for agents. stay tuned for updates 🚀

Get the latest jobs and industry news

AI agents are quickly shifting from AI systems that generate answers to AI systems that take action. In my latest Forbes piece, I explore why governance can’t stay focused on outputs. When agents are coordinating workflows, calling tools, and modifying production systems, the real question becomes: what did the system actually do, and can you prove it? Here’s how I think about it: ➡️ Governance has to move from filtering outputs to controlling actions. Once agents can execute across systems, guardrails need to live in the execution path – not just at the interface. ➡️ The right abstraction is an agentic data plane: a layer that enforces identity, policy, and observability across models, tools, and systems in real time. ➡️ If you can’t reconstruct behavior, you don’t have governance. Full, ordered transcripts of every interaction aren’t optional – they’re the system of record. ➡️ Enterprises will not standardize on one model or vendor anytime soon. Governance has to be consistent across a heterogeneous, multi-cloud, multi-model world. ➡️ Agents should be treated like any other production system: measured not just on uptime, but on correctness, safety, and policy adherence. We’re entering a new phase where AI systems not only assist but also operate. The organizations that succeed won’t just build smarter agents. They’ll build the infrastructure to govern exactly what those agents do. Give the article a read: https://lnkd.in/gjEzDHDi