Agents

Building with Buildprint agents

Buildprint agents help you build faster by letting you ask questions and make changes to your Bubble app in natural language. Here's how to use Agent Chat for common development workflows.

Starting a conversation

Open Agent from the sidebar, then click the + button to start a new thread. Select the project you want to work on from the dropdown - you cannot switch projects mid-conversation.

Each conversation is saved in your thread history. Use the search bar to find past discussions, or archive threads you're done with to keep things organized.

It's recommended to archive conversations when you're done with them. This makes it easy to understand what you're working on.

Debugging with logs

When something breaks, ask the agent to search your logs instead of manually scrolling through Bubble's log viewer.

Example questions:

  • "Find all 500 errors in the last 48 hours"

  • "Show me failed workflow runs for the signup process"

  • "What API calls to Stripe failed yesterday?"

The agent returns relevant log entries with timestamps and context. Follow up with clarifying questions to drill deeper.

Understanding your database and workflows

Ask about your data structure or workflow logic without opening the Bubble editor. Useful when inheriting an app, onboarding teammates, or debugging unexpected behavior.

Example questions:

  • "What tables power the onboarding flow?"

  • "Show me the schema for the User table"

  • "How does the payment workflow work?"

  • "Where do we send confirmation emails?"

The agent explains conditional logic, traces workflow paths, and shows relationships between data types.

Making changes with Allow edits

By default, agents only read your app. To make changes, toggle from View only to Allow edits in the chat composer.

Agents can change anything in your Bubble app, including:

  • Database fields (create, modify, or delete)

  • Workflow steps and conditional logic

  • App settings and configurations

  • API connector endpoints

Test agent changes in a dev version of your app before applying them to production. Always review what the agent suggests before confirming changes.

Example editing requests:

  • "Add a 'last_login' field to the User table"

  • "Create a workflow that sends a welcome email when a user signs up"

  • "Update the checkout flow to handle failed payments"

The agent will explain what it's going to change before making modifications. You can iterate on the plan until it's correct.

Attaching files and Linear issues

Drag files into the composer or click Attach to give the agent additional context - wireframes, specs, error screenshots, or feature requirements.

Link Linear issues with Add attachment → Link Linear issue to connect development work to your task tracker. The agent can reference issue descriptions and comments when planning changes.

Choosing models and reasoning levels

Select your AI model from the dropdown in the message composer. Different models balance speed, capability, and price.

Set reasoning level to None for quick questions, or High when the agent needs to think through complex changes. Higher reasoning is slower but more thorough.

Configure default models and reasoning in Agent → Integrations to avoid selecting every time.

Using Linear integration

Admins can connect Linear in Agent → Integrations. Once connected, mention @Buildprint in Linear issue comments to start an AI conversation with your app's context—useful for triaging bugs or planning features without switching tools.

Configure which Linear teams map to which Buildprint projects, and set default models and permissions for agent responses.

Was this helpful?