Tock
Tock, the open source conversational AI toolkit.
Last verified:
What is Tock?
Tock (The Open Conversation Kit) is a complete and open-source platform to build conversational agents, also known as bots. It is a full-featured Natural Language Understanding (NLU) platform that leverages open technologies like OpenNLP, Stanford CoreNLP, Duckling, Rasa, Spacy, and BERT. Tock does not depend on third-party APIs, allowing users to choose which components to embed and maintain ownership of their conversational data and models.
Key features include Tock Studio, a graphical user interface for building stories and viewing analytics; conversational DSLs for Kotlin, Node.js, Python, and a REST API for any language; built-in connectors for numerous text/voice channels including Messenger, WhatsApp, Google Assistant, Alexa, Twitter, Apple Business Chat, Teams, and Slack; and toolkits for custom Web/Mobile integration with React and Flutter. The platform supports both cloud and on-premise deployments with Docker, and can run embedded bots without Internet connectivity.
Tock is designed for developers, data scientists, and organizations building chatbots for web sites, mobile apps, social networks, and smart speakers. It has been used in production since 2016 by OUI.sncf and growing organizations in energy, banking, and healthcare sectors. The platform source code is available on GitHub under the Apache License version 2.0, running on JVM platforms with Vert.x and MongoDB on the server side.
The platform supports standalone bots or integration with existing systems, multichannel deployments with fine-tuned responses per connector, and IoT use cases where bots run embedded without Internet. Tock Studio is built with Angular in TypeScript, and the FAQ Management feature allows business users without technical conversational knowledge to create and manage question/answer pairs.
Tock pricing
Pricing model: Freemium
Tock is open source under Apache License 2.0 with no licensing costs. The platform source code is freely available on GitHub. Users can self-host on their own infrastructure (cloud or on-premise) at no software cost. A sandbox/demo platform is available for testing at demo.tock.ai. There is no mention of official paid hosting plans from Tock.ai - organizations typically deploy their own instances. The open source nature means costs are limited to infrastructure, hosting, and maintenance rather than software licensing.
Tock pros
- Open source under Apache License 2.0 with no licensing costs
- No dependency on third-party APIs for core functionality
- Full ownership of conversational data and models maintained by users
- Tock Studio graphical interface for building stories without coding
- Built-in NLU platform supporting OpenNLP, Stanford CoreNLP, Duckling, Rasa, Spacy, BERT
- Conversational DSLs available for Kotlin, Node.js, and Python
- REST API enables bot creation in any programming language
- Built-in connectors for Messenger, WhatsApp, Google Assistant, Alexa, Twitter
- Supports Apple Business Chat, Teams, Slack connectors
- React toolkit for easy web integration
- Flutter toolkit for mobile app integration
- Cloud or on-premise deployment options with Docker support
- Embedded bots can run without Internet connectivity for IoT use cases
- FAQ Management menu for business users without technical knowledge
- Production-proven since 2016 by OUI.sncf and other organizations
- Multichannel support with fine-tuned responses per connector
- MongoDB support with DocumentDB alternative available
- Vert.x framework for high-performance server-side operations
- Live demo and sandbox platform available for testing
- Tutorials and user manual for developers available
Tock cons
- Requires JVM platform knowledge for server-side deployment
- Reference language is Kotlin which has smaller community than Python
- Learning curve for conversational concepts like intentions, entities, stories
- NLU accuracy depends on quality and quantity of training sentences
- Minimum 10 varied questions recommended per FAQ for recognition rate
- Node.js framework separate from main Kotlin implementation
- Business users need botUser role to access FAQ Management
- Documentation spread across multiple pages and sections
- Not as versatile as OpenAI bots for general knowledge questions
- Bots need explicit training for domain-specific knowledge
- No built-in free tier hosting mentioned on documentation
- Self-hosting requires infrastructure management
- Docker configurations in separate GitHub repository
- No free trial mentioned for paid hosting options
- Angular-based Studio requires web development knowledge for customization
Frequently asked questions about Tock
What is Tock?
Tock (The Open Conversation Kit) is a complete and open platform to build conversational agents, also known as bots. It is an open-source conversational AI platform that does not depend on third-party APIs, allowing users to maintain ownership of their conversational data and models. The platform source code is available on GitHub under the Apache License version 2.0.
What programming languages does Tock support?
The reference language for Tock is Kotlin, and Tock Studio is built with Angular in TypeScript. However, Tock provides conversational DSLs for Node.js and Python, plus a REST API that allows creating bots and stories using any programming language. The application runs on JVM platforms.
What channels can Tock bots integrate with?
Tock provides built-in connectors for numerous text/voice channels including Messenger, WhatsApp, Google Assistant, Alexa, Twitter, Apple Business Chat, Teams, and Slack. React and Flutter toolkits are also provided for custom Web and Mobile integrations.
How do I get started with Tock?
You can start by reading the tutorial and using the demo/sandbox platform at demo.tock.ai. The user manual for developers is available in the documentation. Tock Studio provides a graphical interface for building stories and analytics without deep coding knowledge.
What is Tock Studio?
Tock Studio is the graphical user interface for building stories and viewing analytics. It is built with Angular in TypeScript and provides menus for Language Understanding (intents and entities), Stories & Answers, FAQ Management, Settings (Applications, Configurations, Namespaces), and security roles.
What is a Story in Tock?
A Story (or Scenario) is the main unit of the Tock conversational framework. It is a functional grouping that allows answering questions on a well-defined subject, generally initiated by a main intention and optionally using secondary intentions. For example, a weather story could ask for location after recognizing the weather intention.
How does FAQ Management work in Tock?
The FAQ Management menu allows creating, modifying, and enriching conversational models with Frequently Asked Questions type questions (Question/Answers with simple text). It is intended for business users not familiar with conversational concepts. Each FAQ requires a minimum of 10 varied questions for adequate recognition rate, and answers are limited to text format with markdown support.
Can Tock be deployed on-premise?
Yes, Tock supports both cloud and on-premise setups, with or without Docker. The platform provides Docker configurations for containerized deployment. Embedded bots can also run without Internet connectivity, making it suitable for Internet of Things use cases.
What NLU technologies does Tock support?
Tock leverages open NLU technologies including OpenNLP, Stanford CoreNLP, Duckling, Rasa, and later supports Spacy and BERT. The platform does not depend directly on these NLU libraries, giving users flexibility to choose which components to embed.
Who is using Tock in production?
Tock has been used in production since 2016 by OUI.sncf (Web/mobile, messaging platforms, smart speakers) and increasingly by organizations in energy, banking, and healthcare sectors. The platform has a growing community as documented on Tock.ai.