ChatBotKit is a platform for building and running conversational AI agents, positioned to let AI agents work continuously in the background, run on schedules and triggers, connect to the tools users already have, and cover every channel.
Purpose The platform is aimed at teams that want to use AI agents to handle requests, inbound leads, and repetitive operational tasks. The official description emphasizes that requests, leads, and the same operational tasks pile up faster than teams can handle them, and agents can take over and keep running.
Main capabilities
- Start from ready-made solution blueprints, or design agents yourself using a visual designer, which the official description says requires no coding.
- Add memory, knowledge, and capabilities to agents in natural language.
- Connect the tools agents should use, such as Slack, CRM, customer service systems, and email.
- Choose the appropriate AI model.
- The platform handles waking agents on schedule, responding to triggers from tools, deploying to every channel, and scaling, which the official description says requires no servers, queues, or on-call staff.
- Provides run observability: view every run and conversation, determine what needs human intervention, and set rules for how agents work.
- Supports approvals before sensitive actions and agents with separated permissions.
- Built-in enterprise-grade security, content filtering, data protection, and compliance capabilities.
- Provides secure built-in connections to existing systems, with all conversations and run records centralized in one history.
- Can control where knowledge is stored and who it serves.
- Can package a runnable system and deploy it again.
How to use it The official path given is "build, run, control": users describe the task, shape the agent, start from a solution blueprint or design their own; add memory, knowledge, and capabilities; connect tools; choose a model; after that, the platform handles scheduling, triggering, channel deployment, and scaling. The official description also mentions "one platform, four entry points," and you can start with a website widget or build a complete agent system, both running on the same hosted platform.
Use cases Officially listed examples include: an operations manager agent (wakes every workday, tracks tasks, follows up on commitments), a WhatsApp business concierge (a permission-separated system: restricted concierge, payments and CRM operator, proactive outreach agent), a proactive Zendesk support agent (monitors the queue and proactively contacts teammates when needed), a ChatGPT campaign manager (reviews campaigns every morning and reports changes), a Telegram personal agent (remembers context and completes tasks), and a website assistant (answers visitors based on your own content and hands work off to other parts of the system). Case studies in customer service, medical charity, and cultural heritage are also mentioned. On channels, it supportsSlack, Discord, WhatsApp, Telegram, and others.
Limitations The materials do not provide specific pricing, free quota, model provider list, deployment regions, SLA, or technical limitations, so none are available for now.
