What is Wit?
Wit is an AI app builder that lets developers add natural language understanding to bots, mobile apps, smart home devices, and wearables. It processes what users say — by voice or text — and converts those inputs into structured data your application can act on. Meta owns and maintains the platform, and it is free to use.
Wit works through a training model. Developers submit example phrases called utterances, assign intents to them, and tag entities to extract specific values. The platform learns from each validated example and improves recognition accuracy over time. As an AI app builder supporting 132 languages, it adapts with every interaction.
Features & Benefits
- Intent Recognition — Identify what a user wants from a spoken or typed phrase, mapped to developer-defined intents with a confidence score.
- Entity Extraction — Pull structured values from utterances, such as temperatures, dates, locations, or custom-defined data points.
- Built-in Intents, Entities, and Traits — Access a library of pre-trained natural language components to reduce training time.
- Utterance Training Interface — Submit and validate example phrases through a visual dashboard to improve model accuracy.
- HTTP API — Query your trained app via API to integrate natural language understanding into any product or service.
- SSML Support — Format speech output for voice-enabled applications using Speech Synthesis Markup Language.
- Multi-platform SDKs — Build with official libraries for Node.js, Python, Ruby, Go, and Unity.
- 132-Language Support — Deploy apps that understand users across more than 132 languages, with models that improve through use.
- Free and Open Platform — Use the full AI app builder at no cost, with no feature restrictions tied to pricing tiers.
Real-World Applications
Developers building messaging bots can use Wit to interpret user input and trigger the right response logic. A support bot may recognize whether a user wants to check an order status or request a refund, then route the request accordingly. Intent recognition and entity extraction handle the language layer, so developers focus on the action logic.
Mobile app developers can add voice or text interaction to any app without building a natural language model from scratch. A fitness app might let users log workouts by speaking naturally, with Wit parsing the activity type, duration, and intensity as structured entities.
Smart home developers can connect Wit to speakers, thermostats, and lighting systems. A voice command like “set the temperature to 68 degrees” gets parsed into an intent and a numeric entity value, which the device controller can act on directly. This AI app builder handles multilingual input, so the same application can serve users across different regions.
Wearable device makers can embed conversational interfaces into products with limited screen space. Wit processes short, context-dependent commands and returns clean structured output, making it practical for devices where typed input is not an option. Teams that need a free AI app builder with broad language coverage and an open HTTP API will find it covers most natural language use cases out of the box.
