The AI support agent for your product.

Kai is the AI customer support agent in Gleap. It answers customer questions from your help center, snippets, files and connected sources, cites the content it used and brings Kai Resolve or your team in when a question needs more. It runs on the chat widget, email, WhatsApp, Instagram, Facebook, Slack and Telegram.

Yes. Kai is the AI agent inside Gleap's customer support platform: shared inbox, help center, channels and the agents that answer, investigate and fix. You can run Kai alone on your widget or as the first level of a full support setup. See the multichannel customer support platform.

An AI agent. A support bot follows scripted flows; Kai retrieves the matching content for each question, writes the answer, cites its sources and can call tools such as API lookups, MCP servers and web search. Scripted steps still exist in Gleap workflows, for example collecting an email before Kai answers.

Snippets, websites, help articles, files (PDF, Word, PowerPoint, Excel, Markdown), API data sources, code repositories, news and changelog posts, Notion pages and YouTube videos. Published help articles are picked up automatically. Tag content to scope what Kai may use; the default scope excludes content tagged internal. See knowledge base software for support.

It says so and, depending on your policy, offers a handover. Kai Resolve investigates the question with the session evidence and your code, or the conversation goes to your inbox with the whole thread. You choose the policy: escalate on its own, ask the customer first, or a standing button in the widget. See Kai Resolve.

Yes. The Talk to our support team button and a request for a human keep working. With Kai Resolve as second level, that request is investigated first and passed to your team when it cannot be settled; without it, the request goes straight to your inbox.

Kai answers customer questions from your knowledge. Kai Resolve investigates the tickets Kai cannot answer and returns a verdict with evidence. Kai Code turns confirmed bugs into pull requests your engineers review. Together they close the loop from question to shipped fix.

Chat widget, email, WhatsApp, Instagram, Facebook, Slack and Telegram. The Let Kai answer workflow step is available on all seven; escalation to Kai Resolve runs on the widget. Source links render as plain links outside the widget.

Yes, for the questions your knowledge covers. Kai handles how-to and account questions from your help center and docs, links feature requests and files bug reports with session details. Commonplace reports that Kai handles roughly 50 to 60 percent of its inbound cases; that is their result, not a benchmark. Read the Commonplace story.

Kai answers first and cites the reply. What it cannot settle follows one path: Kai Resolve investigates, Kai Code fixes confirmed bugs, your team reviews and ships, and the customer hears back in the same conversation. That is self-driving development applied to support.

Kai uses auto routing by default; Pro and Enterprise plans can pick a model from the current catalog. Your knowledge is embedded and stored for retrieval, and customers only receive articles their audience is allowed to see. Security details are in the Trust Center.

Kai Copilot is the internal agent in the ticket composer. Ask it for a draft, similar tickets or the right article; it stages the draft in your composer and sends nothing to the customer. See Kai Copilot.

The Kai report shows questions asked, deflection rate, the share handed to the team, answer rate, positive feedback and the conversation rating, plus the negative feedback worth reading.

Kai is included in every Gleap plan, from Starter up. AI usage is billed separately by tokens and model; Pro and Enterprise add model selection and 25% off AI tokens. See pricing.