#Agent Component
1 messages · Page 1 of 1 (latest)
Hi @compact nova. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
Agent components in Convex are used any time you want AI “agents” with memory, tools, and workflows running in your backend. From the docs, key use cases include:
-
Chatbots and conversational UIs with memory
- Persist user-specific threads and messages so conversations can continue over time and across devices.
- Automatically include prior messages as context in each LLM call. [AI Agents; AI Agent component]
-
Agentic workflows and orchestration
- Build multi-step, long-running workflows that call agents, tools, and other Convex functions (often together with the Workflow component).
- Patterns like dynamic routing between agents, fanning out LLM calls, ReAct-style loops, and workflows that can be paused/resumed. [AI Agents; Complex workflows]
-
Support / customer-service agents
- Define a “support agent” that answers questions, calls tools (e.g. account lookup, ticket creation), and runs inside durable workflows. [AI Agents with memory]
-
Human-in-the-loop / human agents
- Let humans act as agents: save human replies into the same threads, mix LLM and human responses, and decide when a human vs. LLM should answer. [Human agents]
-
RAG-backed assistants
- Combine agents with the RAG component to search custom knowledge bases and inject relevant context into prompts, either directly or via tools. [RAG]
-
Tool-using agents over your data & APIs
- Give agents tools that can read/write Convex data, call external APIs, or trigger other actions, with tool calls and results stored in the same thread history. [Tools; AI Agent component]
-
Multi-agent / collaborative systems