Creating an Agent
Learn how to create an AI agent for your application in Docana to automate tasks and provide intelligent assistance.
Give Agents Website Access
Let agents browse specific sites, including ones behind a login, with stored credentials.
Testing Agents
Write test cases for your agents and catch problems before your users do.
Test Agents in CI
Run agent evals in CI to catch regressions and check changes before publishing.
Insights de conversa
Explore os insights das conversas em uma tabela, adicione classificações e envie relatórios. Os controles de geração e de resumo ficam nas Configurações de insights.
Testes A/B
Compare um agente publicado com uma candidata fixada em conversas reais e depois decida o que publicar.
Público de experimentos
Escolha quem participa de um experimento A/B, mande suas próprias conversas para a variante e meça por cliente em vez de por conversa.
Routines
Run agents on a schedule, on document changes, or from an API call.
Run Agents from Your Systems
Execute an agent from your backend with one API call, and poll for the result.
Organize Agent Projects
Keep agents in git with a project manifest. Review, test, and deploy them like code.
JavaScript Node
Transform workflow data with code: exact math, parsing, and formatting without spending an LLM call.
Webhook Node
Send HTTP requests from a workflow, directly or through a regional relay.
Google Search Grounding
Let an instructions node search Google while it runs, and hand the sources it found to the rest of the workflow.