An LLM (large language model) is a neural network trained on huge amounts of text and code. It understands requests in plain language and generates a response: text, code, a table or a plan of action. Examples include GPT, Claude, Gemini, GigaChat and YandexGPT.
On its own, a model only responds to a prompt. To get it to do real work, you build an AI agent around it and give it tools, data access and memory.
Pachca doesn't lock you into one model: you can connect any agent, cloud or local. The agent works in threads that already hold the context of the discussion, and it talks to Pachca through the API, Pachca CLI and Agent Skills.
For developers
Docs on dev.pachca.com →