Fix a failed deploy

Netlify’s AI capabilities help diagnose and suggest solutions for deploy failures or build errors so you can get back to shipping code.

How these solution suggestions work

When a deployment failure happens, your team will find a Why did it fail? button on the failed deploy summary page. This button always appears on failed deploys unless an Owner has disabled the feature for your team.

If this feature is enabled for your team and you select the Why did it fail? button, the details about the failure are processed by Netlify’s AI systems to generate guidance for how to resolve the issue.

We’re continuously improving the feature to suggest accurate solutions but your team should review them for accuracy.

Selecting Why did it fail? doesn’t consume any credits. If you go on to start an agent run to fix the deploy, that run does consume credits. Learn more in Credit usage for fixing a failed deploy.

Security for solution suggestions

This feature does not use the information processed to train models or store data outside of Netlify’s systems.

The deploy failure details used to generate suggestions includes:

• a subsection of the build log entries in your deploy log at or around the place where the error returns.
• other metadata, such as the name of your site’s framework to provide more accurate results.

Like all Netlify features, usage of this capability is subject to our service agreements referenced in the Terms of Use.

Enable solution suggestions

These AI capabilities are only available if they have been enabled for your team. Once enabled, any Developer or Owner on your team can use this AI capability to generate solutions for resolving deploy failures on all sites they have access to in your team.

Who can turn this feature on or off:

• Owners can enable or disable this feature for their team.
• Developers can enable this feature for their team unless an Owner has disabled this feature for the team. The Why did it fail? button appears to Developers and Owners unless the feature is disabled.
  • If a Developer selects the Why did it fail? button and the feature is not enabled or disabled for the team, then a prompt appears to enable this feature for the team. Until this feature is enabled on the team, selecting this button will not process the deploy failure details or generate suggestions.

As an Owner, to enable deploy failure solution suggestions:

1. For your team, go to Team settings > General > AI enablement
2. Select Configure.
3. Choose Enabled.

Disable solution suggestions

Once disabled, the Why did it fail? button will not appear on any of your team’s sites. Only an Owner can enable deploy diagnostics for all sites in your team after the feature is disabled.

As an Owner, to disable deploy failure solution suggestions:

1. For your team, go to Team settings > General > AI enablement
2. Select Configure.
3. Choose Disabled.

Fix a failed deploy with an agent

Once you have a diagnosis and a suggested solution, you can select Fix with agent to start an agent run that works on the fix for you. You can also start an agent run from any deploy details page by selecting Run AI agent.

Agent Runners have their own requirements, including a Credit-based plan with credits available or an eligible Enterprise plan.

Learn more about how agent runs work in Make changes with Agent Runners.

Credit usage for fixing a failed deploy

Diagnosing a deploy failure and starting an agent run to fix it have different effects on your team’s credit balance.

The following actions don’t consume credits:

• Selecting Why did it fail? to get a diagnosis and a suggested solution. This feature doesn’t generate AI inference usage.

Selecting Fix with agent starts an agent run, and agent runs consume credits through the following usage meters:

• AI inference for the AI model usage during the run. The cost depends on the AI agent, model, and effort level the run uses.
• Compute for the environment the agent works in, measured in GB-hours.
• Web requests and bandwidth for the related traffic.

Publishing the fix applies the credit cost for a production deploy. If that production deploy returns a failing error, it doesn’t consume any credits.

To check what a run cost, expand a task within the agent run to find its credit usage. For more detail on rates and how Netlify calculates costs, check out Pricing for AI features and How agent runs consume credits.

Want to limit Agent Runners?

To help you control costs for Agent Runners, you can set an AI Credit Usage Limit. Once the limit is reached, your team can no longer start agent runs and active agent runs stop. The limit applies to AI inference from Agent Runners only, so AI Gateway usage isn't affected. On Enterprise plans, the limit factors in AI inference from both Agent Runners and AI Gateway — though only agent runs are paused when it's reached. Learn more in our docs on Managing AI features.

More troubleshooting help

• Verified Support Guide on using the log to debug your build process
• Build troubleshooting tips