Agentic AI
Quote
Why is it called "Agentic"?đź”—︎
The term "agent" comes from the Latin "agere," meaning "to do" or "to act." This is precisely what sets agentic AI apart—its ability to act independently in pursuit of defined goals. Unlike generative AI systems that simply respond to queries or generate outputs, agentic AI systems can understand a goal, take initiative, maintain persistent objectives, and adapt their strategies based on real-world feedback. Put simply, an AI agent is a system that uses AI and tools to accomplish actions in order to reach a given goal autonomously.
Throughout our experience, we've found that the best way to understand AI agents is to think about… secret agents! Like James Bond or Jason Bourne, these operatives act autonomously on behalf of their governments, equipped with specialized skills and resources to accomplish specific missions. They don't just analyze situations or make recommendations—they execute. They gather intelligence, make decisions, and take action, persistently working toward their objectives while staying within the boundaries set by their superiors.
Agentic AI operates on the same principle. It doesn't just generate insights—it takes action. It can interact with applications, manipulate data, control hardware, and execute real tasks to achieve specific goals. In fact, an agent can be trained to do anything a human can do on a computer. An agent operates in a continuous loop of planning, reasoning, and execution—learning from each step to refine its approach until the goal is achieved. In essence, it's like having a highly capable assistant who doesn't just know what to do but actually does it—though, as we'll explore later in this book, success depends on providing clear and precise goals and instructions.
— Pascal Bornet, "The Promise (and Limitations) of AI Agents," in Agentic Artificial Intelligence (2024)
Also available on Spotify premium here.
Generative AI vs Agentic AI Comparisonđź”—︎
| Characteristic | Generative AI | Agentic AI |
|---|---|---|
| Core Capability | Generating text, images, code, or music based on learned patterns | Planning, decision-making, multi-step execution without human intervention |
| Memory & Context | Limited memory (short-term context retention, no persistent memory) | Persistent memory (remembers past interactions, adjusts plans accordingly) |
| Autonomy Level | Requires human prompts to generate responses | Operates with minimal human input, executing complex workflows |
| Integration with External Systems | Minimal integration (relies on APIs or tools for external functions) | Deep integration (connects with APIs, databases, physical systems) |
| Learning Ability | Static - learns only through retraining by developers | Evolves - learns from interactions and refines behavior |
| Typical Use Cases | Content creation, summarization, coding assistance, brainstorming | Workflow automation, personal assistants, business operations |
| Business Impact | Enhances efficiency in content-heavy tasks but does not automate workflows: • Average increase speed: 25% faster • Average quality improvement: 40% |
Drives automation, reduces human workload, enhances business scalability: • Time savings: 30-60% • Process acceleration: 40-90% faster |
| Examples | ChatGPT, Claude, Gemini, DALL-E, Midjourney, Copilot | AutoGen, MS Copilot Agent Builder, UiPath Agent Builder, OpenAI Operator, Google Vertex, Crew.ai, Relevance.ai, Agentforce |
Source: Pascal Bornet, "The Promise (and Limitations) of AI Agents," in Agentic Artificial Intelligence (2024)