Agentic work
Also written: Agentic workloads
AI that can autonomously plan, decide, and act to achieve goals.
Agentic work refers to AI systems designed to operate independently. Instead of just responding to a direct command, these AI agents can figure out the steps needed to complete a task, make decisions along the way, and take actions to reach a final goal.
Think of it like hiring an assistant. You tell them the overall objective, like 'organize my files,' and they figure out how to sort them, where to put them, and do the work without you needing to micromanage every single step. Current examples involve AI agents collaborating to update information on a shared platform.
Agentic work involves AI systems capable of autonomous planning, reasoning, and action execution to achieve predefined objectives. These agents can break down complex goals into sub-tasks, monitor progress, adapt to changing circumstances, and interact with environments or tools to fulfill their mission. This contrasts with traditional AI models that primarily perform reactive tasks based on direct input.
The development of agentic capabilities is a key focus, enabling AI to move beyond simple prompt-response interactions towards more proactive and goal-oriented behaviors. This often involves sophisticated internal state management, long-term memory, and tool use.
Frontier AI labs are increasingly developing AI systems that can operate with greater autonomy. Recent advancements focus on enabling these AI agents to perform complex tasks, make decisions, and execute actions independently, moving beyond simple instruction following. This shift is driven by the pursuit of more capable and versatile AI that can handle sophisticated workloads.
- BigQuery Graph measures explained (Why your AI agents get metrics wrong and how measures fix it)Google · Google Cloud Tech on YouTube ·
- spark Gemini 3.6 FlashGoogle · Gemini Models ·
- Last Week in AI #343 - GPT-6, OpenAI’s agents chatted on a wiki, Fable 5.1OpenAI · Last Week in AI ·
- Natural, real-time conversations with AIOpenAI · OpenAI Research ·
- A mystery challenger at the AI frontierGoogle · The Rundown AI ·
- OpenAI President Greg Brockman on GPT-5.5 “Spud,” AI Model Moats, and Cybersecurity RisksOpenAI · Big Technology Podcast ·
- AWS and OpenAI announce multi-year strategic partnershipOpenAI · OpenAI News ·
- Google Gemini
- Google Gemini's work on BigQuery Graph measures and Gemini 3.6 Flash highlights efforts to provide accurate data processing and faster AI for agent tasks.
- OpenAI ChatGPT
- OpenAI is actively developing AI agents capable of autonomous chat and is exploring advanced models like GPT-6 Astra, indicating a strong focus on agentic capabilities.
- Task automation
- This is a broader category where repetitive tasks are automated
- Prompt engineering
- This involves crafting specific inputs to guide AI responses
- Supervised learning
- This is a training method where AI learns from labeled data
The concept of agentic behavior in AI has roots in artificial intelligence research dating back decades, particularly in the field of intelligent agents. The term has gained prominence with recent advancements in large language models enabling more sophisticated autonomous capabilities.
Ask an AI chatbot like ChatGPT to 'plan a weekend trip to a national park for two people, including travel, accommodation, and activities.' Observe how it outlines steps, suggests options, and asks clarifying questions, demonstrating a degree of independent planning to achieve the goal.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- PromptRole-play a student struggling with math for AI tutor simulationSimulate a student's math struggles and uncertainty for AI tutor evaluation. · 5 min
- TemplateReusable workflow templates for healthcare teamsTeams can create and share templates for common workflows like documentation or prior authorizations. · 15 min
- TemplateTemplate for building an AI tool tailored to your workA structured approach to defining and building a custom AI tool for specific business needs. · 15 min
- TemplateEvals CSV prompt list for agent skillsA CSV file template for listing prompts used in evaluating agent skills. · 20 min
- HeuristicDouble coding agent task completion with docs linksProviding coding agents with a documentation link significantly increases their task completion rate. · 2 min
Questions people ask
- What is the difference between an AI agent and a regular AI model?
- A regular AI model typically responds to direct prompts. An AI agent can autonomously plan, make decisions, and take actions to achieve a goal over time.
- Can AI agents make mistakes?
- Yes, AI agents can make mistakes. Their planning and decision-making processes are based on their training data and algorithms, which can lead to errors or suboptimal outcomes.
First seen on Subvolts Nov 3, 2025. Updated Oct 13, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.