Assistants

Assistants in Sila are like GPTs in ChatGPT: reusable, configurable helpers that follow instructions, use selected AI models, and perform tasks across your workspace.

What an assistant is

• A set of instructions and defaults your chats inherit
• A chosen model (OpenAI, Google, Anthropic, Ollama, or any OpenAI‑compatible API)
• Optional tools and data sources it can use
• A target folder for where new conversations are created

Why use assistants

• Consistency: Apply the same style, constraints, and steps every time
• Speed: Start new chats with the right context and model pre‑selected
• Focus: Scope tools and sources to the task at hand
• Reuse: Share assistants across multiple chats and workspaces

Typical uses

• Research assistant that always cites sources and saves results to a folder
• Coding reviewer that follows a specific checklist
• Writer that enforces tone, audience, and structure
• Support bot that searches a chosen knowledge base

How it works in Sila

• Local‑first: Assistants, their instructions, and chats are plain files you own
• Bring your own keys: Use external providers on your terms
• Model‑agnostic: Switch models without losing your assistant’s context
• Workspace‑aware: Organize assistants per workspace and folder

Create assistants, iterate on their instructions, and use them across chats to standardize workflows while keeping full ownership of your data.