agentic workflow
An iterative AI system architecture where a language model engages in planning, reflection, tool use, and revision loops rather than generating output in a single pass, enabling the model to review its own work and improve through multiple iterations.
As defined in What's next for AI agentic workflows ft. Andrew Ng of AI Fund
Where it comes up · 2
- What's next for AI agentic workflows ft. Andrew Ng of AI Fund9 claims
- 3 Ways To Get Better At AI Right Now4 claims
Most important claims · 10 of 13Search all claims for “agentic workflow” →
- On the HumanEval coding benchmark, GPT-3.5 with zero-shot prompting achieves 48% accuracy, GPT-4 achieves 67% accuracy, but GPT-3.5 wrapped in an agentic workflow actually outperforms GPT-4, and an agentic workflow around GPT-4 performs even better.What's next for AI agentic workflows ft. Andrew Ng of AI Fund
- This agentic workflow result has significant consequences for how practitioners approach building AI applications, suggesting that architectural pattern and iteration strategy matter as much or more than raw model capability.What's next for AI agentic workflows ft. Andrew Ng of AI Fund
- Agentic workflows with language models involve iterative processes where the model writes an outline, conducts web research, drafts content, reviews its own work, identifies revision needs, and revises iteratively—delivering remarkably better results than non-agentic approaches.What's next for AI agentic workflows ft. Andrew Ng of AI Fund
- Generating more tokens quickly from a slightly lower-quality LM might deliver comparable results to generating fewer tokens slowly from a higher-quality LM in agentic workflows, suggesting quality-speed tradeoffs are different in iterative architectures.What's next for AI agentic workflows ft. Andrew Ng of AI Fund
- The typical non-agentic workflow with language models involves typing a prompt and generating an answer in a single pass, analogous to asking a person to write an essay without ever using backspace.What's next for AI agentic workflows ft. Andrew Ng of AI Fund
- Building agentic workflows manually with tools like N8N, Zapier, Plum, or Lindy gives you a much better understanding of how agents actually work under the hood, and this leg up will carry forward even when using more abstracted consumer agents.3 Ways To Get Better At AI Right Now
- Agentic workflow tools become more powerful when plugged into other external systems, but they should give you the ability to control when and how the agent has access to those systems through authorization and confirmation toggles.3 Ways To Get Better At AI Right Now
- Fast token generation is important for agentic workflows because agents iterate repeatedly, with the LM generating tokens for the agent to read, and the ability to generate tokens much faster than humans read enables more efficient looping.What's next for AI agentic workflows ft. Andrew Ng of AI Fund
- If someone does the three recommended activities—using O3 as a strategic collaborator for a week, vibe-coding with a tool like Lovable, and building agentic workflows—they will become better than 99.999% of the population at actually getting value out of AI.3 Ways To Get Better At AI Right Now
- A critical challenge with agentic workflows is human impatience with response latency: modern users expect instant responses (sub-second, like web search), but agentic workflows require patience for extended execution times (minutes to hours), which conflicts with human nature and management instincts.What's next for AI agentic workflows ft. Andrew Ng of AI Fund