Make AI Browser Agent
Make AI Browser Agent is currently available to select users in closed beta. Functionality, available AI models, and pricing may change while the app is in beta.
Make AI Browser Agent opens a live web browser and interacts with a website for you. Describe a task in plain language, and the agent completes it on the website the same way you would: navigating pages, clicking buttons, typing, submitting forms, or extracting data.
Use this app when a task requires interacting with a browser, such as filling in forms on sites with no API, or automating recurring tasks that are behind two-factor authentication (2FA).
Make AI Browser Agent modules
You can use the following module to build your scenarios.
Browse a website and complete a task
Opens a web browser and completes a task on a website: navigating pages, filling in forms, clicking buttons, or extracting data, based on a plain-language instruction.
Field | Description |
|---|---|
Instruction | The task in plain language. Tell the agent where to start (for example, "Go to example.com and…") and what to do. |
Browser context | If the website requires signing in, select a saved browser context or add a new one. |
Save browser context changes | When enabled, changes made during the run, such as new cookies, updated sign-in state, and website data, are saved in the selected browser context for future runs. When disabled, the run uses the saved context without updating it. |
Allowed domains | Restrict browsing to specific domains. Enter example.com to allow that domain and its subdomains, or *.example.com to allow only subdomains. Requests to any other domain are blocked. Leave empty to allow all domains. |
Max steps | Maximum number of steps the agent takes to complete the task. Defaults to 20. |
Timeout (minutes) | Maximum time the agent runs before the task is cancelled. Defaults to 10 minutes. |
AI model | The underlying AI model that runs the task. |
Response format | How the agent returns its answer. Text is free-form output, Data structure defines fields inline for the agent to fill in, and Data structure (reusable) uses an existing Make data structure. |
Response structure | Shown when Response format is a data structure. Defines the fields the agent returns. |
Viewport width/height (px) | Browser window size in pixels. Leave empty to use the provider default. |
Record session video | Disabled by default. Records the full session, including any data displayed on screen, and stores it with the browser provider. Enable only for tasks that exclude sensitive data. |
Log session | Disabled by default. Logs session activity. Logs may capture sensitive data displayed or entered during the run. |
Record session video and Log session capture whatever is visible or entered on screen. Don't enable them for tasks that involve sensitive data.
Output
Field | Description |
|---|---|
result | The agent's text answer. |
jsonResponse | The structured result, when Response format is a data structure. |
creditsUsed | Credits used in the run. |
tokenUsage | AI model and token counts used for the run. |
durationMs | How long the run took. |
sessionReplay | Link to a video of the session, if Record session video was enabled. |
steps | A per-step summary of what the agent did. |
Browser context
Browser context is data captured during sign-in that Make AI Browser Agent reuses in future runs. It includes cookies, local and session storage, service workers, and saved form data, and excludes cached images, scripts, and fonts.
Create browser contexts for websites that require authentication or show popups. Once created, you and your teammates can reuse them in scenarios.
If a website shows popups like cookie banners, close them while creating or updating the browser context so they won't appear again in future runs.
Create a browser context
You can create a browser context from Credentials > Browser contexts, or directly from a Make AI Browser Agent module while building a scenario.
To create a browser context:
Choose where to start:
- From Credentials > Browser contexts: In the left sidebar, go to Credentials, then the Browser contexts page. In the top right, click Create browser context.
- From the Make AI Browser Agent module: In the module settings, toggle on Advanced settings. In Browser context, click Add.
In Name, name the browser context to identify it later.
In the browser session that opens, sign in to the website and complete any two-factor authentication (2FA) or verification steps.
Click Save & finish.
You've now created a new browser context.
Save browser context changes
By default, a run using a browser context reads the context's data, but doesn't change it.
Once you create a browser context, enable Save browser context changes in the module settings to have that run's cookies and session state write back into the browser context afterward. This option is useful when the task involves signing in again, or you expect the website to rotate session state.
Manage browser contexts
From the Browser contexts list, you can update, rename, and delete browser contexts.
Update a browser context
Update a browser context when the website signs out a saved session, for example, after a password change, an expired token, or forced re-authentication.
To update a browser context:
In Credentials > Browser contexts, next to a browser context, click the three-dot menu.
Click Update.
In the browser, sign in again.
Click Save & finish.
You've now updated a browser context. This change is automatically reflected in any scenarios using this browser context.
Rename a browser context
To rename a browser context:
In Credentials > Browser contexts, next to a browser context, click the three-dot menu.
Click Rename.
Click Save & finish.
You've now renamed a browser context.
Delete a browser context
Before deleting a browser context, check which scenarios use it; deleting a browser context breaks any modules that still use it.
To delete a browser context:
In Credentials > Browser contexts, next to a browser context, click the three-dot menu.
Click Delete.
Click Save & finish.
You've now deleted a browser context.
Credit usage
Make AI Browser Agent consumes credits based on AI tokens and browser session time:
- AI tokens: Vary based on the task and underlying AI model. For token-to-credit conversion rates for each AI model, see the All other models table in Make's AI Provider.
- Browser session time: Minutes spent in the browser session. Each minute equals approximately 2.5 credits.
Expect lower credit usage when Make reuses data from scenario runs with the same instruction and AI model. When this happens, Make replays a previous run instead of calling an AI model. Data extraction tasks are the exception; they always require calling the AI model.
For tips on reducing credits, see the next section.
Best practices
Follow these tips to reduce credit usage and write effective instructions when using Make AI Browser Agent:
- Write your instructions as step-by-step actions instead of one broad goal. The more specific the steps, the more reliably the agent completes them.
- Focus on one website and task at a time.
- Set conservative limits in Max steps and Timeout.
- Use the cheapest AI model for simple browsing tasks and more expensive ones for higher-reasoning tasks.