What can you do with our APIs?
Join our discord here to get help and see what people are building!👾
Voiceflow lets teams build, test, and launch AI agent (also referred to as a 'project' or 'assistant') to any channel, for any use case.
Voiceflow offers a suite of APIs to launch and host AI Agents, as well as build custom tools, interfaces and workflows on-top of Voiceflow.
Popular APIs include:
- Dialog Manager API: launch your Agent with hosted dialog state management
- Knowledge Management API: manage Knowledge Base data with a CRUD API
- Knowledge Query API: query your managed Knowledge Base
- Project API: query content from your Voiceflow Agent design represented in JSON
- Analytics API: retrieve Agent usage data
- Transcripts API retrieve end-user conversation transcripts
- NLU API train and run inference on VFNLU
These APIs can be used together with Voiceflow's Agent building platform to create custom solutions.
Popular solutions include:
- Single Source of Truth: Use the Voiceflow Agent Builder as a single-source of truth for Agent designs, programmatically retrieving structured design content (JSON) with the Project API
- Build and Launch: build low-code with the Agent builder, deploy with the Dialog API
- Managed Knowledge Base: use the Knowledge Base Management API and Knowledge Base Query API to act as a managed Knowledge Base service
We have a few built-in integrations and let you deploy to a channel like Webchat, WhatsApp, SMS, and Microsoft Teams quickly with minimal code. These are fast but are limited in customizability.
Using our Dialog API directly gives you the flexibility to launch to any channel. These code samples and others on our github provide a starting point to launch to a custom channel.
Voiceflow is a collaborative tool that features a powerful collaborative low-code AI agent building platform. Agents built with Voiceflow are represented by structured JSON files called .VF files.
The below Voiceflow Agent design is an example of what is structured using the .VF JSON file.
You can find Voiceflow's Agent design platform documentation here.
Agents built on Voiceflow offer a Dialog API which manages conversation state at runtime.
This enables developers to offload the mundane work of managing dialog state machines to the designer who creates and maintains the conversation logic with Voiceflow's graphic design interface. Developers can use the Dialog API for production use cases, or custom interface prototypes, where Voiceflow is hidden in the background.
Private cloud or premises Dialog Manager deployments are available upon request. The Dialog API can use Voiceflow's NLU, or your own 3rd party or custom NLU to resolve intents.
Agents built on Voiceflow act as a headless CMS as the Agents content and structure can be pulled at any moment using the Project API.
For teams with large custom Agents on legacy NLU platforms, the design-development hand-off can be a pain due to design work being done with spreadsheets and misc flowcharts.
Voiceflow allows non-technical team-members to design, test and iterate on Agent content while developers can pull the structured JSON content.
This means that developers don't need to re-create an Agent from scratch and can instead focus on custom integrations.
You can add Custom Actions to Voiceflow's design canvas for your non-technical team-members to use and represent any functionality not natively represented in Voiceflow.
Once an agent has been built on Voiceflow, there are two options for deployment:
- Dialog Manager API: run the Agent's state management, and NLU through Voiceflow's Dialog Manager API
- Project API: pull content from the Agent design by API, effectively using Voiceflow as a CMS
Alternatively, Voiceflow offers one-click integrations with many channels that allow for no or low-code deployment options. These are called Channel Integrations.
Updated 14 days ago
More details on our Dialog Manager API: