On-demand Webinar

Derisking Releases in the AI Era | Harness Resource

Shipping faster usually fails for one simple reason: teams try to increase release velocity by taking on more risk. Progressive delivery acts as a technical bridge that allows high-performing teams to decouple deployment from release, which enables speed and experimentation with less harm and a whole lot more user feedback.

The 2025 DORA report found that AI tends to increase pull request (PR) size by 154%, expanding the blast radius of any release. Breaking work into small, manageable chunks reduces that risk, while progressive delivery practices can reduce your blast radius. Both are essential to manage deployments when unpredictable AI elements are involved.

In this panel, hear how high-performing teams are actively following platform engineering best practices, reducing blast radius, and turning every rollout into a learning loop.

Join the panel and learn:

How to increase deployment frequency without increasing risk with progressive delivery
How to build an experimentation rhythm so teams can validate ideas with user behavior
How to shift from “big-bang” releases to lower-risk, smaller daytime releases

Published

March 10, 2026

Guide on its way

Check your inbox — your playbook is ready.

You're all set

Check your inbox — your download is on the way.

What you'll learn

Key Takeaways

Feature Flags Create Psychological Safety

Modernizing tech stacks can inadvertently slow teams down if developers fear breaking production. Using feature flags provides a safety net that reduces release risk and accelerates development velocity.

Decouple Deployments From Feature Releases

Progressive delivery allows teams to separate the deployment of code from the actual release of software. This enables organizations to turn on specific functions safely at their own pace.

Target Rollouts Based on Risk Tolerance

Segment users based on their willingness to accept risk before rolling out new features. Offering early access to opt-in customers prevents breaking workflows for those who require stability.

Manage User Fatigue from AI-Driven Changes

The rapid pace of AI development can lead to excessive user interface changes that frustrate customers. Teams must prioritize A/B tests carefully to avoid exhausting users with constant updates.

Leverage AI for Experiment Analysis

Beyond generating code, AI serves as a powerful investigative tool for analyzing experiment data. It helps engineers uncover hidden causes when performance metrics do not move as expected.

Rely on Quantitative Data for Feedback

While direct customer conversations provide immediate anecdotal feedback, quantitative measurements are essential for faster operational decisions. This ensures greater confidence when evaluating the impact of a release.