Skip to main content
Amazon Nova Act is a Python SDK for building and deploying AI agents that automate browser-based workflows at scale. Define your workflows using natural language prompts combined with Python code, and let Nova Act handle the complexity of browser automation.

​
What is Amazon Nova Act?

Amazon Nova Act is available as an AWS service to build and manage fleets of reliable AI agents for automating production UI workflows at scale. Nova Act completes repetitive UI workflows in the browser and escalates to a human supervisor when appropriate.

AI-powered automation

Use natural language to control browsers and complete complex tasks

Production-ready

Deploy workflows to AWS AgentCore Runtime for scalable automation

Human-in-the-loop

Enable human supervision with approval workflows and UI takeover

Structured extraction

Extract typed data from web pages using Pydantic schemas

​
Key features

​
Browser automation with natural language

Control web browsers using AI-powered agents that understand natural language instructions. No need to write complex selectors or wait conditions. 
from nova_act import NovaAct

with NovaAct(starting_page="https://example.com") as nova:
    nova.act("Find flights from Boston to Seattle on Feb 22nd")

​
Workflow orchestration

Combine Python code with natural language prompts to build sophisticated automation workflows. 
from nova_act import NovaAct, Workflow

@workflow(
    workflow_definition_name="flight-booking",
    model_id="nova-act-latest"
)
def book_flight():
    with NovaAct(starting_page="https://example.com") as nova:
        nova.act("Search for flights from Boston to Seattle")
        nova.act("Select the cheapest option")
        price = nova.act_get("Return the price").response
        return price

​
Data extraction

Extract structured data from web pages using Pydantic schemas for type-safe responses. 
from pydantic import BaseModel

class FlightData(BaseModel):
    price: float
    departure_time: str

result = nova.act_get(
    "Find the cheapest flight and return its details",
    schema=FlightData.model_json_schema()
)
flight = FlightData.model_validate(result.parsed_response)

​
Human-in-the-loop

Enable human supervision for approval workflows, CAPTCHA solving, and complex decision-making. 
from nova_act.tools.human.interface.human_input_callback import (
    HumanInputCallbacksBase, ApprovalResponse
)

class MyCallbacks(HumanInputCallbacksBase):
    def approve(self, message: str) -> ApprovalResponse:
        # Show approval UI to human
        return ApprovalResponse(approved=True)

with NovaAct(human_input_callbacks=MyCallbacks()) as nova:
    nova.act("Complete the purchase")

​
Use cases

Quality assurance

Automate testing of web applications and user workflows

Form filling

Fill out repetitive forms with data from various sources

Data extraction

Extract structured data from websites at scale

E-commerce automation

Automate product searches, comparisons, and purchases

​
Getting started

1

Install the SDK

pip install nova-act
2

Set up authentication

Get your API key from nova.amazon.com/act
export NOVA_ACT_API_KEY="your_api_key"
3

Run your first workflow

See the quickstart guide for a complete example

​
Next steps

Quickstart

Get up and running in minutes

Core concepts

Learn about workflows and browser automation

API reference

Explore the complete API documentation

Deployment guide

Deploy to production with AWS

​
Learn more

Build docs developers (and LLMs) love

Get started for freeTalk to us