What is a chatbot?
A chatbot is an interface that takes in user input, runs the underlying engine — either an Agent or a pipeline — and gives back a response, while maintaining a persistent conversation history. Once you build a chatbot, you choose how people reach it: a shareable link, an embedded widget on your website, a Slack bot, a WhatsApp or SMS number, or a programmatic API call. Every chatbot has three parts:- An Agent or pipeline as the engine. An Agent autonomously decides which tools to call and in what order based on the user’s message. A pipeline runs a fixed sequence of nodes you define. Either way, when a user sends a message, VectorShift runs the underlying engine and returns the output as the chatbot’s reply.
- Conversation history. For Agent-based chatbots, conversation history is managed automatically. For pipeline-based chatbots, include a Chat Memory node in your pipeline to pass prior messages into the LLM so it can reference what was said earlier.
- A deployment channel. Once the chatbot is built, you choose how people reach it: a shareable link, an embedded widget on your website, a Slack bot, a WhatsApp or SMS number, or a programmatic API call.
Two types of chatbots
VectorShift supports two distinct types of chatbots, each designed for different levels of autonomy and control.| Aspect | Agentic chatbot | Workflow chatbot |
|---|---|---|
| Definition | An autonomous chatbot that decides which tools to call and in what order, based on the user’s message | A chatbot backed by a fixed pipeline where you define every step of the logic explicitly |
| Nature | Autonomous flow: the agent selects tools, retrieves data, and chains actions on its own | Deliberate flow: the pipeline runs the same sequence of nodes every time |
| Best for | Open-ended tasks where the user’s intent varies widely (scheduling, multi-system lookups, dynamic Q&A) | Predictable tasks where every conversation follows a known pattern (order support, FAQ, document search) |
| How you build it | Select an agent as the source, add tools, and let the agent figure out the execution path | Build a pipeline with Input, LLM, Chat Memory, Knowledge Base, and Output nodes, then connect it to a chatbot |
| Deployment | Same five channels: share link, embed, Slack, WhatsApp/SMS, API | Same five channels: share link, embed, Slack, WhatsApp/SMS, API |
Ready to build?
Build an Agentic Chatbot
Create an autonomous chatbot powered by an Agent that picks its own tools — the recommended way to get started
Build a Workflow Chatbot
Create a chatbot backed by a step-by-step pipeline where you control every stage of the logic
Already have a chatbot?
Chat Assistant vs Website Chatbot
Choose how your chatbot appears to your users — full-page assistant or embedded widget
Deploy to Slack
Let your team chat with the bot directly in your Slack workspace
Deploy via WhatsApp / SMS
Connect your chatbot to WhatsApp or SMS through Twilio
Chat via API
Integrate chatbot functionality into your own application using the REST API
Analytics
Track usage, review conversations, and export data to improve performance