Getting Started with Feluda

Getting Started with Feluda

This guide will help you complete your first session in Feluda.

You will learn how to:

• download and open Feluda;
• understand the main areas of the application;
• connect an AI model;
• send your first message in Workbench; and
• choose what to explore next.

You do not need to know how to code. Start with one AI model and one simple task. You can add workflows, Genes, tools, and automation later.

What you need

Before you begin, make sure you have:

• a Windows, macOS, or Linux computer;
• the Feluda desktop application; and
• access to either a supported cloud AI provider or a compatible local AI model.

A cloud AI provider usually requires an account, an internet connection, and an API key from that provider.

A local AI model runs on your own computer through compatible software such as Ollama or LM Studio. This option can be useful when you prefer local processing or want to work without an online AI provider.

You only need one working model to get started.

Step 1: Download Feluda

Visit the Feluda download page and choose the version for your operating system.

Feluda is available for:

• Windows;
• macOS; and
• Linux.

Follow the installation guide for your operating system. When installation is complete, open Feluda.

If your computer shows a security notice, read it carefully and follow the instructions in the matching Feluda installation guide.

Step 2: Open the Dashboard

The Dashboard is the first place you see when Feluda opens.

It gives you access to the main parts of the application:

You do not need to configure every area during your first session.

Begin with AI Providers, then move to Workbench.

Step 3: Choose your first AI model

Feluda needs access to an AI model before it can answer a message or use an AI step in a workflow.

You can begin with either a cloud model or a local model.

Option 1: Use a cloud AI provider

Choose this option when you already have access to a supported online AI service.

You will usually need an API key from the provider.

To connect it:

1. Open AI Providers from the Feluda sidebar.
2. Choose the provider you want to use.
3. Add the requested provider details.
4. Save the provider.
5. Confirm that its available models appear in Feluda.

Treat an API key like a password. Do not place it inside a chat message, prompt, document, or workflow instruction.

Use the provider setup area or Secrets when Feluda asks you to store a private value.

Option 2: Use a local AI model

Choose this option when you want the model to run on your own computer.

Before connecting it to Feluda:

1. Install a compatible local model application, such as Ollama or LM Studio.
2. Download a model that is suitable for your computer.
3. Make sure the local model service is running.
4. Open AI Providers in Feluda.
5. Choose the matching local provider.
6. Ask Feluda to find the available models.
7. Select the model you want to use.

Local models differ in speed, quality, and computer requirements. Begin with a smaller model if you are unsure what your computer can run comfortably.

Step 4: Open Workbench

Workbench is the easiest place to confirm that your model is connected correctly.

Open Workbench from the sidebar.

Then:

1. Choose your provider.
2. Choose one of its available models.
3. Enter a short message.
4. Send it.
5. Review the response.

You can begin with:

Explain Feluda in three short sentences for someone using it for the
first time.

If the model answers, your first provider is ready to use.

Step 5: Try a useful task

After the first test, try a task that resembles your everyday work.

For example:

Read the meeting notes below.

Return:
1. a short summary;
2. the decisions that were made; and
3. a list of action items with their owners.

Meeting notes:
[Paste your example notes here.]

Clear instructions help the model understand what you expect.

A useful request normally includes:

• the task;
• the information to use;
• the format you want; and
• any limits or rules that matter.

Use non-sensitive sample information while you are learning how the selected model behaves.

Step 6: Review the result

Do not assume that an AI response is correct simply because it sounds confident.

Check whether the response:

• used the information you provided;
• followed your requested format;
• left out important details;
• added anything that was not in the source; and
• is suitable for its intended use.

Change your instruction and try again when the result is unclear or incomplete.

This testing process will also help you later if you turn the task into a workflow.

Step 7: Decide what to do next

Once Workbench is working, choose your next step based on what you want to achieve.

Continue using Workbench

Stay in Workbench when you want to explore an idea, ask one-time questions, compare models, or improve an instruction through conversation.

Build a workflow in Studio

Open Studio when you have a task with clear steps that you expect to repeat.

A simple first workflow could:

1. receive a piece of text;
2. ask an AI model to summarise it; and
3. return the summary.

Keep your first workflow small. Add more steps only after the basic version works.

Explore Genes

Explore Genes when you want to add a useful tool, prompt, workflow, or other focused capability to Feluda.

Read a Gene's description and requirements before adding it. Test it with non-sensitive information before using it for important work.

Run a saved workflow

Use RunFlows when you already have a workflow that is ready to test or use.

Provide the requested input, start the flow, and review the result before relying on it.

A good first-session path

You do not need to explore every feature immediately.

A simple first session looks like this:

1. Download and open Feluda.
2. Connect one AI provider.
3. Open Workbench.
4. Send a short test message.
5. Try one useful task.
6. Review the answer.
7. Decide whether to continue in Workbench or build a small workflow.

Completing these steps gives you a working Feluda setup and a clear starting point for the rest of the documentation.

Keep your first setup safe

Follow these basic practices while setting up Feluda:

• keep API keys and passwords out of prompts;
• use sample information while testing;
• check which provider and model are selected;
• review enabled tools before using them;
• read the requirements of a Gene before adding it; and
• review important AI-generated results yourself.

When you use an external AI provider or tool, the information needed for the task may be sent to that service. Review the service's own terms and privacy information before using sensitive data.

If something does not work

If no model appears in Workbench, return to AI Providers and confirm that the provider was saved and that models are available.

For a cloud provider, check that:

• the provider details are correct;
• the API key is active;
• your internet connection is working; and
• your provider account can use the selected model.

For a local provider, check that:

• the local model application is open;
• its model service is running;
• at least one model has been downloaded; and
• Feluda can find the available models.

If Feluda itself does not open or install correctly, use the dedicated installation guide for your operating system.

You are ready to continue

Once a model responds in Workbench, your basic Feluda setup is complete.

You can now learn how to write clearer AI instructions, create your first workflow in Studio, install a Gene, or connect another AI provider.

Start with one small task, test it carefully, and expand your setup as your needs become clearer.