Skip to content
Tavora

Introduction

Tavora is an Agentic Intelligence Platform as a Service that lets SaaS companies add AI capabilities — document understanding, semantic search, and agent reasoning — without building infrastructure.

What Tavora provides

Section titled “What Tavora provides”
  • Document ingestion and RAG — Upload documents (PDF, text, HTML), automatically chunk, embed, and index them for semantic search powered by pgvector.
  • Chat completions — OpenAI-compatible chat API with retrieval-augmented generation built in.
  • Agent reasoning — Autonomous agent runtime with multi-step reasoning, tool use, and function calling via Google Gemini.
  • Skills and MCP servers — Extend agent capabilities with prompt skills, webhook skills, JavaScript module skills (custom code in the Goja sandbox), and MCP server integrations.
  • Scheduled runs — Cron-based agent execution for recurring tasks.
  • Evals — Evaluation framework to measure agent quality and track regressions.
  • Playground — Interactive experimentation UI for chat, document upload/search, and agent runs — try everything before writing code.
  • Studio — Time-travel debugger for agent execution traces with AI-powered fix analysis.

How it works

Section titled “How it works”

Tavora is a backend-to-backend service. You integrate it into your own backend using the Go SDK or REST API, scoped to a workspace (your tenant in the platform).

sequenceDiagram
    participant User as Your Users
    participant App as Your Backend
    participant SDK as Tavora SDK
    participant API as Tavora API
    participant AI as AI Pipeline

    User->>App: Request
    App->>SDK: SDK call (API key scoped to workspace)
    SDK->>API: REST API request
    API->>AI: Document processing / RAG / Agent reasoning
    AI-->>API: Result
    API-->>SDK: JSON response
    SDK-->>App: Typed Go structs
    App-->>User: Response

Each workspace has its own documents, conversations, agents, skills, and configuration. Most deployments put Tavora behind their own backend (server-to-server SDK calls), but browser apps can also talk to Tavora directly using a short-lived workspace API key minted from the logged-in user’s session — see Browser-side chat for that pattern.

Next steps

Section titled “Next steps”
  • Quickstart — Make your first API call in 5 minutes
  • Playground — Try chat, documents, and agents interactively
  • Skills — Extend agents with custom JavaScript, prompts, and webhooks
  • Go SDK — Integrate Tavora into your Go backend
  • REST API — Full API reference