WEBVTT

00:00:00.200 --> 00:00:01.566
Hello everyone!

00:00:01.766 --> 00:00:06.066
In this tutorial, I’ll show you how to connect your
AI subscriptions to WebBrain.

00:00:06.100 --> 00:00:15.166
Currently, you can use WebBrain with your own API keys or local AI
tools like Ollama, llama.cpp, or LM Studio.

00:00:15.200 --> 00:00:22.233
However, you can also integrate your existing subscriptions,
including Claude, ChatGPT, and Gemini.

00:00:22.233 --> 00:00:26.766
To get started, you'll need a simple application
called Easy CLI Proxy API.

00:00:26.966 --> 00:00:29.200
I'll provide the link for you.

00:00:29.433 --> 00:00:35.566
It features a clean graphical interface, and the installation process
is standard, so we'll dive right in.

00:00:35.566 --> 00:00:38.266
In the app, navigate to the OAuth tab.

00:00:38.300 --> 00:00:42.766
Even if the runtime is stopped, this section remains active.

00:00:42.766 --> 00:00:45.833
From here, you can sign in to your various subscriptions.

00:00:45.866 --> 00:00:48.233
Let's use Google Antigravity as an example.

00:00:48.266 --> 00:00:52.966
A browser window will open for you to complete the
authentication process.

00:00:52.966 --> 00:00:58.766
Once finished, you'll see a success message, and you can return to
the Easy CLI Proxy app.

00:00:58.800 --> 00:01:03.966
Now, back on the home screen, you'll find your local API endpoint and
authentication key.

00:01:03.966 --> 00:01:09.733
Simply copy the API URL for the model you want to use, along with the
authentication key.

00:01:09.733 --> 00:01:15.100
Next, switch over to WebBrain and open the settings
menu to configure your provider.

00:01:15.100 --> 00:01:20.400
Under the Providers tab, search for the “Local
OpenAI-compatible Proxy” option.

00:01:20.400 --> 00:01:25.500
Paste the Server URL and your API Key into the designated fields.

00:01:25.533 --> 00:01:29.633
Click “Load Models” to see your available options.

00:01:29.633 --> 00:01:33.166
In this case, we'll select Gemini 3.7 Flash.

00:01:33.200 --> 00:01:40.533
For Gemini, you can set a large context window—up to one million
tokens—though we'll use a standard range for this test.

00:01:40.533 --> 00:01:46.166
Ensure “Model Supports Vision” is checked and set the Prompt Tier to
“Full” for the best performance.

00:01:46.166 --> 00:01:49.433
Save your settings and run a connection test.

00:01:49.466 --> 00:01:52.300
You should see a green success message.

00:01:52.500 --> 00:01:53.866
Let's test it out!

00:01:54.100 --> 00:01:57.933
We'll start with “Ask Mode” on Bluesky to see if
the model responds correctly.

00:02:01.166 --> 00:02:07.400
Finally, we'll test “Act Mode” by having WebBrain draft and publish a
post automatically.

00:02:17.066 --> 00:02:18.233
And that's it!