Agentic AI
AI systems where an LLM plans and executes multi-step tasks by calling external tools, accessing files, browsing, and adjusting its own approach based on results.
Agentic AI moves beyond chat — the LLM is the policy network of an autonomous loop. The agent receives a goal, plans steps, calls tools (search, code execution, DB query, browser, file system), observes results, and re-plans. Modern agent SDKs (Anthropic's Claude Agent SDK, OpenAI's Agents SDK, LangGraph, Mastra) standardize the loop, tool definition, memory, and human-in-the-loop checkpoints. Production agent fleets need monitoring, cost guardrails, and clear evaluation harnesses.
Related terms
Anthropic's official Python and TypeScript SDK for building production agents with Claude — handles the tool-use loop, hooks, and slash commands.
An AI pattern that retrieves relevant documents from a vector database and injects them into the LLM prompt — so the model can answer from custom knowledge it was not trained on.
A capability of modern LLMs where the model emits structured JSON requesting that the host program execute a named tool with arguments — used to give the LLM internet, code execution, database, or any external action.
A neural network with billions of parameters trained on broad text corpora to predict and generate language — the engine behind ChatGPT, Claude, and Gemini.
Read more on the blog
Need this built into a real product?
Viprasol Tech ships production code for everything defined here — MT4/MT5 EAs, AI agents, B2B SaaS, AWS architecture.
Send a brief →