AI Agent
An LLM-driven system that plans, calls tools, and executes multi-step tasks autonomously.
An AI agent is an LLM given a goal, a set of tools (APIs, databases, code execution), and a loop: observe, decide, act, repeat. Unlike a chatbot that responds once, an agent can spend minutes or hours pursuing an objective — booking travel, resolving a support ticket, or reconciling invoices.
Modern agent frameworks (LangGraph, CrewAI, OpenAI Agents SDK, Anthropic's Claude Agent SDK) provide the plumbing for tool calling, memory, and human-in-the-loop approval gates.
The engineering challenge with agents is not intelligence — it is reliability, cost control, and observability. Production agents need step-level tracing, budget caps, and structured escape hatches to humans.
Key points
- Reliability drops as steps compound — 95% per step becomes 60% over 10 steps
- Tool definitions and error messages are the highest-leverage design surface
- Budget and timeout limits are mandatory in production
- Human-in-the-loop for high-stakes actions (payments, emails, deletions)
Common use cases
Frequently asked
Related terms
A neural network trained on massive text corpora to understand and generate human-like language.
The practice of designing model inputs to produce reliable, task-specific outputs.
An open standard that lets LLMs discover and call external tools, data, and services in a uniform way.
A structured pipeline where an LLM makes decisions at defined branch points, but the overall flow is deterministic.
Applying AI Agent in your business?
We help enterprises design, ship, and govern AI systems end-to-end.