Docs

What Research Agent is, first sign-in, service setup, and your first research.

The deep research pipeline: search, download, extraction, report, verification.

Seven providers (including Yandex results via JsonSeo), fallbacks, search sessions, relevance filtering.

Five fetch methods — from a direct request to an undetectable browser, working around blocks, conversion to Markdown.

Chunking (splitting long documents into readable pieces), batch extraction to TSV/JSON, batches, tokens and cost.

The document tree, the editor, source preview, Edit with AI.

Generating ideas, plans, and interfaces for development agents.

plan → act → observe → reflect cycles on three levels.

A ready-made marketing and sales knowledge base, structured so you can extend it.

Agent chats in the interface, calling the agent from a document, the agent API.

Search providers, fetch methods, LLM services, priorities, copying settings.

Docker Compose, environment variables, first sign-in.

Prepare tasks in advance and the agent works through them on its own: AutoSend, Max active chats, a recurring prompt generator.

Dictate a task straight into the agent chat — no typing needed; speech is recognized by the browser or by the configured agent’s engine.