Auto-GPT is an application that uses GPT-4 to achieve your goals. Tell it what you want to achieve, and it will try to find out how to do it by searching the internet.
You will learn how to install and run Auto-GPT on Windows.
(See this guide for installing Auto-GPT on Mac.)
Contents
What is Auto-GPT?
Auto-GPT is an open-source program exploring the utility of the GPT-4 language model. You tell it some goals, and It uses GPT-4 to come up with a step-by-step plan to achieve your goals.
It can perform actions such as searching the internet to find relevant information. It will come up with the next steps based on the result of the first step. It will go on and go until your set goals are achieved.
Many goals would require Auto-GPT to perform multiple actions to complete. You can pre-approve it to run on its own. So the name Auto.
Auto-GPT can:
- Search the internet
- Send emails
- Write files
- Write codes
In this guide, we will only enable it to search the internet.
Systems requirement
Auto-GPT uses OpenAI’s API to access GPT-4. It doesn’t need a powerful PC to run.
You should have Windows 10 or above.
Install Auto-GPT on Windows
Step 1: Install Python
Install Python 3.10 if you don’t have it already.
- Visit the Python 3.10 page on Mircosoft Store.
- Click Get in Store App. (You may need to grant permission to open the Store App)
- In the Store App, click Get to install.
Step 2: Install Git
Next, install Git if you don’t have it on your Windows system already.
Visit git’s download page.
Select 64-bit Git for Windows, or use this download link.
Run the downloaded installer to install Git.
Step 3: Clone the Auto-GPT repository
Press the Windows key on your keyboard. (Usually, it is on the bottom-most row, on the left of the space bar.)
Type “powershell” and press enter.

It should launch the PowerShell app. You will use this app to install and run Auto-GPT.

For simplicity, this guide installs Auto-GPT in your home folder. Feel free to change the installation location if you prefer a different one.
In the PowerShell App, run the following command (i.e., copy and paste the text below and press Enter) to go to your home folder.
cd ~
Run the following command to clone the Auto-GPT repository.
git clone https://github.com/Significant-Gravitas/Auto-GPT.git
You should see something like the one below when it is done.

Now you have the Auto-GPT codes copied to your local Windows computer.
Run the following command to go into the newly created Auto-GPT directory.
cd 'Auto-GPT'
Since the code is constantly under development, you can optionally use the stable branch, which is stable and working.
git checkout stable
Step 4: Install the required packages
Run the following command to install the required packages.
python -m pip install -r requirements.txt
Step 5: Create an OpenAI API key
You will need to have an OpenAI account to create the API key. You will be prompted to sign up for one if you don’t have one already.
Open this link: https://platform.openai.com/account/api-keys
Click the “+ Create new secret key” button.

Name the key “Auto-GPT”.

After clicking the button “Create secret key“, you will see the secret key. This is the only time you can see it. Copy the key and paste it into a safe document.
Step 6: Put your API key in Auto-GPT
Now go back to the PowerShell App.
You should be in the Auto-GPT folder. If not, run
cd ~/Auto-GPT
Copy the config file from the template.
cp .env.template .env
A. new file .env
should be created.
Open File Explorer and go to the Auto-GPT location by putting the following in the location bar and pressing Enter.
%userprofile%/Auto-GPT

Find the file .env
. Right-click the file and select Open with. Select Notepad.
The file .env
should now be opened with the Notepad app.

Find the line
OPENAI_API_KEY=your-openai-api-key
Change it to include your API key. It should look like
OPENAI_API_KEY=sk-XXXXXXXXXXXXXXXXX
Save the file and exit the Notepad app.
Step 7: Setup a paid OpenAI account
You will need to set up a paid account at OpenAI in order to use the GPT API.
Go to the billing page: https://platform.openai.com/account/billing/overview
Click Set up paid account.

You have completed the necessary setup after putting in the way to pay them.
You can optionally set a monthly usage limit on the same page. Put down a $10 hard limit and a $5 soft limit.
Step 8: Install Chrome browser
Auto-GPT uses Chrome to browse the internet by default. Download and install Chrome here.
Run Auto-GPT
Open the PowerShell App.
Go to the Auto-GPT folder.
cd ~/Auto-GPT
Start Auto-GPT by running the following command.
python -m autogpt
You should see the following prompt.

Now, name your AI.
Next, you will be prompted to describe your AI’s role. The role should be relevant to the goals you are going to put it later.
For example, I want to ask Auto-GPT to find the best gaming laptop for under $1000.
In AI’s role, enter an AI designed to find the best electronic products autonomously.
Next, it will ask you to put down up to 5 goals. You don’t need to come up with 5. Let’s just use one in your first Auto-GPT project. E.g.
Goal 1: Find the best gaming laptop for under $1000.
When you are prompted for Goal 2, press Enter to finish.
Then Auto-GPT will start to run.

It will tell you its plan and the reasoning behind it.
Enter y to accept the plan.
It’s capable of searching the internet and reading web pages to find the answer for you.

You can keep entering y to approve its next plan.
Or y-5 to pre-approve the next 5 plans.
Enter n when you are done.
Run options
If you have problems running Auto-GPT, run in debug mode to see the error message.
python -m autogpt --debug
You will see information that would normally be hidden, such as whether you are running GPT-3.5 or 4.
Updating Auto-GPT
Auto-GPT evolves quickly. It won’t be the same next week.
To update Auto-GPT, open the PowerShell App.
Go to the Auto-GPT folder
cd ~/Auto-GPT
Update Auto-GPT with the command
git pull
You’d think there could be no such restriction on iPhones
and iPads, since Apple and Google aren’t the best of buddies (or
possibly they are).
Why do I keep getting an error after entering the API key from the website https://platform.openai.com/account/api-keys:SYSTEM: Command write_to_file returned: Error: File has already been updated.
Error: Reached rate limit
read more here: https://docs.agpt.co/setup/#getting-an-api-key
Why do I still get an error after entering the API key? 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
Error: Reached rate limit
can read more here: https://docs.agpt.co/setup/#getting-an-api-key
Waiting 8.0 seconds…
JOCKETERGPT THOUGHTS: To accomplish the goal of writing the generated jokes to a text file, I can use the ‘write_to_file’ command.
REASONING: The ‘write_to_file’ command allows me to write the jokes to a text file, which is exactly what I need to do.
PLAN:
– Use the ‘write_to_file’ command to write the jokes to a text file named ‘JocketerGPT_results.txt’.
CRITICISM:
SPEAK: I will use the ‘write_to_file’ command to save the jokes to a text file.
NEXT ACTION: COMMAND = write_to_file ARGUMENTS = {‘filename’: ‘JocketerGPT_results.txt’, ‘text’: ”}
SY