Ollama & embeddings

The Ollama & embeddings section of Settings configures a local Ollama backend and selects the embedding model your workspace uses. Embeddings turn text into vectors, which power every semantic feature in Chatty.

Use this section when you want to run models locally with Ollama, or when you need to set or change the system embedding model.

Local Ollama backend

Point Chatty at a running Ollama instance to use local models. Once configured, Ollama serves both chat/completion models and the embedding model.

Embedding model

Choose the system embedding model (for example, nomic-embed-text). This model is used whenever text needs to be vectorized.

If you change the embedding model, you must re-vectorize existing content so old and new vectors are comparable — different models produce vectors of different shapes and meanings.

1. Open Settings and go to Ollama & embeddings.

2. Configure the Ollama backend (its host/endpoint), and make sure the embedding model is pulled on that instance.

3. Select the embedding model to use system-wide.

4. If you changed the model, trigger a re-vectorization so existing content is re-embedded.

Caution

Re-vectorizing recomputes embeddings for all affected content and can take a while on large workspaces. Plan to run it after any change to the embedding model.

What embeddings power

The system embedding model drives every semantic feature:

• Semantic context — finding relevant context items by meaning for RAG.
• VectorDB Query nodes — similarity search in graphs; see the VectorDB Query node.
• Episodic and Semantic agent memory — how agents recall relevant past content; see Memory strategies.

Memory strategies How agents remember across turns, including embedding-backed memory.