Build Your Free AI Search Engine in 2024 | Beat Perplexity

As we navigate the information-rich landscape of 2024, AI-powered search engines have become indispensable tools for efficient and intelligent information retrieval. These advanced systems have evolved from simple keyword matching to complex, context-aware platforms that understand and anticipate user needs.

Recently, OpenAI made waves by launching their own version of Perplexity called Search GPT. This strategic move indicates a shift in the AI industry: when third-party companies package and offer services using OpenAI’s models, the tech giant prefers to step in and offer such services directly. This development has not only shaken up the market but also opened doors for innovative alternatives in the AI-powered search engine space.

The Limitations of Perplexity

Perplexity, a popular AI search engine, claims to perform extensive work to generate results, using this as justification for their $20 user fee. However, upon closer examination, their core functionality primarily combines GPT-4 with Google search results. While this approach yields impressive results, it comes with limitations:

1. Lack of transparency in the search process
2. Potential for biased results based on training data
3. Limited customization options for specific user needs
4. Dependency on external APIs, which can lead to downtime or slow responses

While some users argue that nothing can beat Perplexity’s performance, OpenAI’s direct entry into this market with Search GPT has likely unsettled Perplexity’s investors and opened the door for alternatives.

Introducing a Powerful Local Perplexity Alternative

Today, I’ll guide you through setting up a local Perplexity alternative on your computer using Llama 3.1 and Perplexica. This solution offers several advantages:

• Free to use
• Completely local, ensuring privacy and reducing latency
• Customizable to your specific needs
• Independent of external API limitations

Our local setup is remarkably similar to Perplexity in functionality but gives you full control over the AI search engine’s operation.

Setup Options

The setup process is straightforward, with options to integrate with both Ollama and Groq. We’ll explore three configuration options, each catering to different needs and computational resources:

Option	Model	Pros	Cons
1. Llama 3.1 8B with Ollama	Smaller, faster	Lower resource requirements, quicker responses	Less powerful, may struggle with complex queries
2. Groq’s Llama 3.1 70B	Balanced performance	Good for most use cases, free API (with limits)	Requires internet connection, potential rate limiting
3. Llama 3.1 405B	Most powerful	Best performance for complex tasks	Highest resource requirements, may be slower

Let’s dive into the setup process for each option.

Setting Up with Ollama and Llama 3.1 8B

Step 1: Install Ollama

1. Visit the Ollama website and download the appropriate version for your operating system.
2. Install Ollama following the on-screen instructions.

Why this step? Ollama provides a user-friendly interface for managing and running large language models locally.

Step 2: Install Llama 3.1 Model

1. Navigate to the Models page in Ollama.
2. Select Llama 3.1 and copy the installation command: ollama run llama3.1
3. Paste the command into your terminal to begin the model installation.

This step downloads and sets up the core language model that will power your AI search engine.

4. Once installed, test the model by sending a message in the chat interface.

Troubleshooting Tip: If you encounter a “command not found” error, ensure Ollama is properly installed and added to your system’s PATH.

Step 3: Install Embedding Model

Run the following command in your terminal to install the embedding model:

ollama pull nomic embed text

The embedding model is crucial for understanding and representing text in a format that the AI can process efficiently.

Step 4: Install Docker

1. Download and install Docker from the official website.
2. Follow the on-screen instructions to complete the installation.

Docker is necessary for creating a consistent environment to run Perplexica across different systems.

Step 5: Install Perplexica

1. Search for “Perplexica” on Google and click on the GitHub link.

2. Scroll down the GitHub page and copy the provided command.

3. Open your terminal and paste the command to clone the repository to your computer.

This step downloads the Perplexica software, which acts as the interface between you and the AI model.

Step 6: Configure Perplexica

1. Open the cloned folder in VS Code or your preferred text editor.
2. Rename the sample.config.toml file to config.toml.

3. In the terminal, navigate to the cloned folder and run the compose command to create and start a Docker container:

docker-compose up -d

This configuration step ensures Perplexica is set up correctly to work with your local environment.

Step 7: Set Up the Web Interface

1. Open port 3000 in your browser to access the Perplexica interface.
2. If you encounter an error, go to the settings and add the correct URL to the AI API base URL (typically http://localhost:11434).

3. In the settings, select Llama 3.1 as the model and choose “Ollama” as the embedding provider with “nomic embed text” selected.

Troubleshooting Tip: If the web interface doesn’t load, check that Docker is running and that the Perplexica container started successfully.

Diverse Search Capabilities

Like Perplexity, Perplexica boasts a wide range of search functionalities, catering to various user needs:

• Reddit Search: Easily find discussions, opinions, and user-generated content from Reddit.
• YouTube Search: Quickly locate relevant videos on YouTube.
• Other Specialized Searches: Perplexica supports searches across various platforms and databases, providing a comprehensive search experience.

These diverse search options allow users to access a wide range of content types and sources, making Perplexica a versatile tool for information retrieval.

Collaborative Editor

Perplexica features a collaborative editor switch, although its exact functionality isn’t fully clear from the current information. This feature likely allows multiple users to work together on search queries or results, enhancing teamwork and knowledge sharing.

Search History

One of the most useful features of Perplexica is the ability to view your past search records. This functionality offers several benefits:

1. Quick Access to Previous Searches: Easily revisit and continue previous research or queries.
2. Track Research Progress: Monitor your search patterns and the evolution of your queries over time.
3. Improved Productivity: Quickly pick up where you left off without having to remember or retype previous searches.

To access your search history:

1. Navigate to the options menu in Perplexica.
2. Look for the search history section.
3. Browse through your past searches, organized chronologically.

This feature not only saves time but also helps users maintain continuity in their research or information-gathering processes.

By combining these user-centric features with its powerful AI search capabilities, Perplexica offers a comprehensive and user-friendly search experience that rivals and potentially surpasses traditional search engines and other AI-powered alternatives.

Using Perplexica with Groq

For users who prefer not to run the LLM locally but still want to use Perplexica, Groq offers a compelling alternative. Groq provides free API access (with some rate limits) to the Llama 3.1 8B and 70B models.

Steps to Configure Groq:

1. Sign up for a Groq Cloud account at cloud.groq.com.
2. Create an API key in the API Keys section of your account dashboard.

3. In Perplexica settings, enter your Groq API key.

4. Change the provider to Groq and select the Llama 3.1 model (either 8B or 70B).

Performance Comparison:

• Local Setup: Lower latency, complete privacy, no usage limits
• Groq-powered: No local resource usage, potentially more powerful models, subject to API rate limits

Setting Up with Llama 3.1 405B Model

For those requiring maximum performance and willing to dedicate more computational resources, the Llama 3.1 405B model offers state-of-the-art capabilities. This larger model excels at complex reasoning, nuanced language understanding, and generating more detailed and accurate responses.

You can set up this powerful model using Together AI, which offers some free credits to get started.

Steps to Configure Together AI:

1. Sign up for a Together AI account at www.together.ai.
2. Copy your API key from the account settings.

3. In Perplexica settings, change the provider to “Custom OpenAI”.
4. Enter the following details:

• Base URL: https://api.together.xyz/v1
• Model Name: togethercomputer/llama-3.1-405b
• API Key: Your Together AI API key

Advantages of the 405B Model:

• Superior performance on complex queries and tasks
• More nuanced understanding of context and intent
• Improved ability to generate detailed, coherent responses

Use Cases for the 405B Model:

• Advanced research and data analysis
• Complex problem-solving and strategic planning
• Sophisticated content generation and summarization

Conclusion

By following these steps, you’ve now set up your own Perplexity-like AI search engine using Llama 3.1, completely free and locally hosted. This powerful alternative offers functionality similar to Perplexity, including:

• Generating articles, images, and videos
• Answering follow-up questions with context awareness
• Providing search capabilities for platforms like Reddit and YouTube

As AI technology continues to advance, personal AI search engines like the one you’ve just built are poised to become increasingly powerful and indispensable tools. They offer a glimpse into a future where information retrieval and processing are not only more efficient but also more personalized and privacy-conscious.