Chat Thing

What it is

Chat Thing helps businesses create AI customer support agents trained on their documentation, FAQs and website content. The AI agent can be deployed to a website, phone line, Discord, Slack, Telegram, WhatsApp or via API. It learns from connected data sources and answers customer questions automatically 24/7, while allowing human team members to take over complex conversations when needed. The platform includes analytics to track customer inquiries, conversation management, and supports over 50 languages.

Quick answers

What is a data source?

Data sources are the way you connect your bot to your data. Chat Thing currently supports importing data from Notion, web pages, text file formats, and YouTube videos.

What is a message token?

Message tokens are used each time you send a message to a bot and receive a response. Each message uses between 1,000-4,000 message tokens depending on the question and the bot's response. Roughly 200,000 message tokens equals about 100 messages.

Can my bot hand off to a real person?

Yes, on the Standard plan and above. On Standard and up, a customer can ask for a human, the conversation appears in your dashboard, a team member can claim it and chat live, then hand back to the bot. The Basic plan includes the Talk to a Human power-up but without live takeover. The Free plan does not include escalation.

Is my data used to train AI models?

No. Chat Thing never trains on your data. Your content, data sources, and customers' conversations are never used to train AI models. For chat messages, they are not even stored by the AI provider.

Which LLM models are supported?

By default bots use GPT-4o Mini. Standard, Pro, and Enterprise tier subscribers can use advanced LLM models such as GPT-5 from providers including OpenAI, Google, Cohere, Claude, and Mistral.