Date & time

17:00

Two-thirds of vendors expect to offer self-hosted deployment options going forward. Almost none of them have solved the operational problem underneath it.

Your enterprise customer just asked: “can you run this in our environment?” Behind the question is regulated data that can’t cross borders, datasets too costly to egress into a vendor’s cloud, and AI workloads that need to run next to the customer’s data rather than yours. As more of what you ship is AI-powered, that question stops being an edge case and starts being the default ask.

The request gets agreed by sales. The operational load falls to engineering. When incident response and upgrades happen across a growing number of separate environments, ownership of that work isn’t always clearly assigned. What starts as a single self-hosted build tends to multiply, and each one needs the access and infrastructure layer that turns a one-off deployment into something that can be operated at scale.

On the panel: engineering leaders who’ve built and operated Bring Your Own Cloud (BYOC) deployments. You’ll hear what’s actually driving these requests, including the AI-specific pressure to keep models and data co-located, and how the engineering teams who’ve solved this keep it from becoming an ongoing maintenance burden.

You’ll learn how to:

• Launch a self-hosted deployment option without turning it into a second product your team has to maintain forever
• Keep feature parity and secure access in place as one deployment turns into dozens
• Manage the operational problems of running software in someone else’s cloud: upgrades, on-call, and incident response in accounts where you don’t hold the keys
• Read a data-residency requirement the way the buyer’s security team does