canonical: https://jentic.com/apis/dante-ai.com/dante-ai # Dante AI API Dante AI publishes an official OpenAPI specification covering its full platform; the specification Jentic publishes here is a curated, agent-optimized subset, narrowed to the knowledge base and chat operations agents actually call and kept validated and agent-ready. Dante AI lets builders create custom chatbots grounded in their own knowledge bases assembled from files, URLs, or text. The 35-endpoint API covers knowledge base creation, conversations, messages, personalities, models, and text-to-speech, secured with bearer token authentication. It is suited to embedding branded chat assistants in websites, support flows, and internal tools. ## For AI agents Build and query knowledge-base-grounded chatbots - create knowledge bases from files or URLs, run conversations, and synthesise speech with Dante AI. ## Scope Does not handle raw model fine-tuning, image generation, or vector database hosting - use for Dante's knowledge-base-grounded chatbots, conversations, and text-to-speech only. ## Capabilities - Build knowledge bases from uploaded files, URLs, or mixed sources - Search and retrieve content within an existing knowledge base - Run multi-turn conversations grounded in a chosen knowledge base and personality - Configure chatbot personalities to control tone, instructions, and model choice - Generate text-to-speech audio from chatbot responses - Authenticate calls with a bearer token in the Authorization header ## Use cases ### Documentation-Grounded Support Chatbot Spin up a support chatbot grounded in product documentation without building a retrieval pipeline. The POST `/knowledge-bases/url` and POST `/knowledge-bases/files` endpoints ingest content, then conversations and messages endpoints run grounded chat against it. Suitable for SaaS support, internal help desks, and product-specific Q and A widgets. Example prompt: Create a knowledge base from the product docs URL, attach it to a personality named 'Support Assistant', and run a conversation answering a user's question grounded in the indexed content. ### Internal Knowledge Search Build an internal knowledge search interface backed by a Dante knowledge base populated from policy documents and wikis. The GET `/knowledge-bases/{kb_id}/search` endpoint returns relevant passages for a user query without running a full chat session. Useful when a team wants retrieval results without conversational framing. Example prompt: Search the policy knowledge base for 'remote work expense reimbursement' and return the top three matching passages with their source references. ### Voice-Enabled Chat Experiences Pair Dante's grounded chat with text-to-speech to deliver voice-first interfaces in mobile apps, kiosks, or accessibility tools. The Text-to-Speech endpoints turn each chatbot reply into audio with a chosen voice. Suitable for hands-free help surfaces or audio learning experiences. Example prompt: Run a conversation turn in the assistant knowledge base, then call the text-to-speech endpoint to generate an audio file of the reply for playback in the mobile app. ### Multi-Persona Chatbot Platforms Operate multiple chat personas - for example, a sales bot and a support bot - that share underlying knowledge bases but differ in tone and instructions. The /personalities endpoints configure each persona with its own model and system prompts. Useful for vendors offering multi-bot platforms to their customers. Example prompt: Create two personalities - 'Sales Concierge' and 'Support Helper' - both pointing to the shared product knowledge base, and run a conversation turn against each to validate tone. ### Agent-Built Chatbots via Jentic AI agents that assemble custom assistants for downstream users can call Dante AI through Jentic without storing each customer's bearer token. The agent searches Jentic for the relevant intent, loads the schema, and creates knowledge bases or runs conversations on the user's behalf. Credentials remain in your Jentic One instance. Example prompt: Use Jentic to search 'create a knowledge base from urls', load POST `/knowledge-bases/url`, and execute it with the user's documentation URLs to provision their chatbot. ## Key endpoints | Method | Path | Description | | --- | --- | --- | | POST | `/knowledge-bases/url` | Create a knowledge base from URLs | | POST | `/knowledge-bases/files` | Create a knowledge base from uploaded files | | GET | `/knowledge-bases/{kb_id}/search` | Search a knowledge base | | GET | `/knowledge-bases/{kb_id}` | Get knowledge base details | | DELETE | `/knowledge-bases/{kb_id}` | Delete a knowledge base | | PATCH | `/knowledge-bases/{kb_id}` | Update a knowledge base | | POST | `/knowledge-bases/{kb_id}/files` | Add files to a knowledge base | ## Key resources - **Knowledge Bases** — Create, update, search, and delete knowledge bases from files or URLs - **Conversations** — Run multi-turn chat sessions grounded in a knowledge base - **Messages** — Send and read messages within a conversation - **Personalities** — Configure chatbot tone, instructions, and model - **Models** — List models available to the account - **Text-to-Speech** — Generate spoken audio from text replies ## Why Jentic - **Setup:** Wiring Dante AI by hand means learning its bearer auth against api.dante-ai.com and building the upload, search, and retry plumbing for knowledge bases yourself. Through Jentic you install once, import the Dante AI API from the API Directory, store the token once, and your agent calls it. - **Permission scoping:** Dante AI puts the knowledge base id in the URL path (`/knowledge-bases/{kb_id}/...`), so a rule can pin your agent to one knowledge base for searching and adding files. You choose the operations it may call, so deleting a knowledge base is not included unless you add it. - **Credential handling:** Your Dante AI bearer token is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context. - **Discovery method:** Agents search Jentic by intent such as 'create a knowledge base from a URL' or 'search a Dante knowledge base', and Jentic returns the matching operation with its input schema so the agent calls the right endpoint without browsing the api.dante-ai.com docs. ## Related APIs - **OpenAI API** — OpenAI provides general-purpose language models; Dante AI provides packaged knowledge-base-grounded chatbots. - **Anthropic Messages API** — Anthropic's Messages API exposes Claude models directly; Dante adds knowledge base ingestion and persona management. - **ElevenLabs API** — ElevenLabs offers higher-fidelity TTS voices that can replace or augment Dante's built-in text-to-speech. ## FAQ ### Does Dante AI publish an official OpenAPI specification? Yes. Dante AI serves an official OpenAPI 3.0.2 document at `https://api.dante-ai.com/openapi.json`, covering its full platform. The specification Jentic publishes here is a curated subset of that surface, scoped to the 35 knowledge base, conversation, message, personality, model, and text-to-speech operations agents call most, and validated against the live API. Get started with Jentic One, the self-hosted execution layer. ### What authentication does the Dante AI API use? The Dante AI API uses bearer token authentication via the bearerAuth scheme - the token is sent in the Authorization header. Through Jentic the bearer token is held in the vault and never enters the agent's context. ### Can I create a knowledge base from URLs with the Dante AI API? Yes. POST `/knowledge-bases/url` accepts a list of URLs and creates a knowledge base from their content. POST `/knowledge-bases/all` combines URL and file inputs in one call. ### What are the rate limits for the Dante AI API? The OpenAPI spec does not declare explicit numeric limits. Apply backoff on 429 responses and batch knowledge base updates rather than calling PATCH per file when ingesting many sources. ### How do I run a grounded chat through Jentic? Run pip install jentic, then search Jentic for 'send a message to a dante chatbot', load the conversation message endpoint schema, and execute it with the conversation ID and user message. Jentic injects the bearer token at call time. ### Does the Dante AI API support text-to-speech? Yes. The Text-to-Speech endpoints generate audio from text replies, useful for voice-first interfaces or accessibility-focused chat surfaces. ### Can I limit what my agent is allowed to do with the Dante AI API? Yes. Because you run Jentic One yourself, your own rules decide which Dante AI operations and credentials the agent may use. Since Dante AI puts the knowledge base id in the URL path, such as `/knowledge-bases/{kb_id}/search`, you can pin the agent to a single knowledge base for searching content and adding files. You also choose the exact operations it may call, so a destructive action like DELETE `/knowledge-bases/{kb_id}` stays off limits unless you explicitly allow it.