Raven FAQs

Question 1

What is Raven and what is its primary purpose?

Accepted Answer

Raven is an LLM-powered agent designed to manage your personal memories and facts. It serves as a personal memory server, helping you organize, store, and recall diverse information like facts, feedback, projects, and references with high accuracy.

Question 2

How does Raven handle memory storage and retrieval without embeddings?

Accepted Answer

Raven uniquely uses 'context stuffing,' loading all stored memories directly into the LLM's large context window for native semantic matching. All data is stored in inspectable and version-controllable YAML and Markdown files, ensuring transparency and ease of management.

Question 3

What kind of information can I store and manage with Raven?

Accepted Answer

Raven's intelligent `remember` function automatically categorizes and stores various types of information, including queryable facts (e.g., job details, preferences) in YAML, and feedback, project notes, or external references in Markdown files.

Question 4

Is Raven compatible with different Large Language Models?

Accepted Answer

Yes, Raven is highly flexible and supports any OpenAI-compatible LLM provider. This includes OpenAI's own models, OpenRouter, Ollama, LM Studio, or any other provider that adheres to the OpenAI API standard, allowing for cost-effective and customizable operation.

Question 5

What are the advantages of Raven's approach over traditional RAG pipelines for personal memory?

Accepted Answer

For personal memory, Raven's approach of leveraging large LLM context windows eliminates the need for complex RAG infrastructure and vector databases. This results in zero retrieval errors, transparent file-based storage, reduced operational overhead, and simplified, direct access to your information.

Raven

Raven

Key Features

Use Cases

Key Features

Use Cases