agentic workflow
In plain wordsA multi-step process in which AI models draft, check, use tools and revise their work over several rounds, instead of answering in one shot; often the steps are laid out by developers in advance.
- 1
Iterative LLM processes
A process in which language modelslarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code. reach a result through a series of steps, each able to call toolstool, n.: Any outside capability an AI model is allowed to use — a web search, a calculator, a code runner, a database or an app — described to the model so it can ask for it by name., inspect earlier output and improve on it, rather than in a single pass from prompt to final answer. Andrew Ng, who popularized the phrase in 2024 and contrasted it with zero-shot use, grouped these workflows into four design patterns: reflection (self-critique and revision), tool usetool use, n.: An AI model's ability to reach beyond its own knowledge by operating outside tools — searching the web, running code, using a calculator or an app — and folding the results into its work., planning (task decompositiontask decomposition, n.: Breaking a big, vague job into smaller, concrete steps that can each be done and checked — the way a recipe splits a meal into stages.) and multi-agent collaboration. He illustrated their value with a coding result: on HumanEvalHumanEval, prop. n.: A set of 164 small programming exercises, each described in plain English and checked by hidden tests, used to see whether an AI can write working Python code., GPT-3.5 scored 48.1% zero-shot and GPT-4 67.0%, while GPT-3.5 wrapped in an iterative agent loop reached up to 95.1%.
We kept it as an agentic workflow — classify, draft, check — because a free-roaming agent wasn't worth the risk.
- 2
Workflows versus agents
In the narrower usage of Anthropic's 'Building effective agents' (2024), a system in which models and tools are 'orchestrated through predefined code paths', as opposed to an AI agentAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done. that directs its own process. The named workflow patterns are prompt chainingprompt chaining, n.: A way of getting a big job done with a relay of AI requests: each handles one stage and passes its result to the next — pull out the facts, write, polish — with automatic checks between stages., routing (classifying an input and sending it to a specialized prompt or model), parallelization (splitting a task into independent sections, or sampling several answers and voting), the orchestrator–worker patternorchestrator–worker pattern, n.: A way of organizing AI teamwork, like a newsroom: an editor AI decides which stories a big assignment needs, sends reporter AIs to cover them at the same time, and turns what they file into one report. and the evaluator–optimizerevaluator–optimizer, n.: A pairing in which one AI produces a draft and another AI grades it and suggests fixes, repeating the cycle until the work is good enough. loop. Workflows trade flexibility for predictability, cost control and easier testing, and are recommended whenever a task's steps can be known in advance.
Origin
Andrew Ng promoted 'agentic workflows' in a talk at Sequoia Capital's AI Ascent and a series of letters in The Batch beginning 20 March 2024. Erik Schluntz and Barry Zhang's 'Building effective agents' (Anthropic, December 2024) drew the influential line between workflows and agents and named the standard patterns.
Usage note
Overlaps with 'compound AI system' (Matei Zaharia and colleagues, 2024), a broader term for applications that combine several models, retrievers and tools. Many production systems are hybrids: fixed workflows with agentic steps inside.
Related
- prompt chainingprompt chaining, n.: A way of getting a big job done with a relay of AI requests: each handles one stage and passes its result to the next — pull out the facts, write, polish — with automatic checks between stages.
- evaluator–optimizerevaluator–optimizer, n.: A pairing in which one AI produces a draft and another AI grades it and suggests fixes, repeating the cycle until the work is good enough.
- orchestrator–worker patternorchestrator–worker pattern, n.: A way of organizing AI teamwork, like a newsroom: an editor AI decides which stories a big assignment needs, sends reporter AIs to cover them at the same time, and turns what they file into one report.
- AI agentAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done.
- agentic AIagentic AI, n.: AI built to pursue goals on its own — planning, taking actions and adjusting along the way — instead of only responding to one prompt at a time. 'Agentic' describes how much of this independence a system has.
- orchestrationorchestration, n.: Coordinating several AI calls, tools or agents so each does its part in the right order and their results are combined into one finished piece of work, as a conductor directs an orchestra.
- model routingmodel routing, n.: Automatically choosing which AI model should answer each request, a cheap quick one for easy questions and a bigger slower one for hard ones, to save money without hurting quality.
- AI agentAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done.
Sources
- Ng, A. (2024). How Agents Can Improve LLM Performance. The Batch (DeepLearning.AI), 20 March 2024. deeplearning.ai/the-batch/how-agents-can-improve-llm-performance
- Schluntz, E. & Zhang, B. (2024). Building effective agents. Anthropic. anthropic.com/research/building-effective-agents
Mentioned in
agentic AIagentic AI, n.: AI built to pursue goals on its own — planning, taking actions and adjusting along the way — instead of only responding to one prompt at a time. 'Agentic' describes how much of this independence a system has., AI agentAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done., agent frameworkagent framework, n.: A software library that gives developers ready-made parts for building AI agents — model connections, tool handling, memory, multi-agent coordination and logging — so they need not write everything from scratch., orchestrationorchestration, n.: Coordinating several AI calls, tools or agents so each does its part in the right order and their results are combined into one finished piece of work, as a conductor directs an orchestra., orchestrator–worker patternorchestrator–worker pattern, n.: A way of organizing AI teamwork, like a newsroom: an editor AI decides which stories a big assignment needs, sends reporter AIs to cover them at the same time, and turns what they file into one report., plan-and-executeplan-and-execute, n.: An agent design that first writes out a complete step-by-step plan for a task, then carries out the steps one by one, revising the plan only when something goes wrong., prompt chainingprompt chaining, n.: A way of getting a big job done with a relay of AI requests: each handles one stage and passes its result to the next — pull out the facts, write, polish — with automatic checks between stages., robotic process automationrobotic process automation, n.: Software 'robots' that copy the clicks and typing a person would do in office computer programs — such as moving data between forms — to automate repetitive paperwork., task decompositiontask decomposition, n.: Breaking a big, vague job into smaller, concrete steps that can each be done and checked — the way a recipe splits a meal into stages., augmented language modelaugmented language model, n.: A language model hooked up to extras — a search engine, a calculator, a memory store or other tools — so it can do things a model working only from its training cannot., compound AI systemcompound AI system, n.: An AI product built from several working parts — models, search tools, code and checks — designed to work together, rather than a single model answering on its own., evaluator–optimizerevaluator–optimizer, n.: A pairing in which one AI produces a draft and another AI grades it and suggests fixes, repeating the cycle until the work is good enough., LangGraphLangGraph, prop. n.: A free software library for building AI agents as flowcharts of steps that share a memory, so developers control exactly how an agent moves between steps, pauses and resumes.