Reader's Guide
Although this primer can be read in its entirety without much technical background beyond a basic understanding of Large Language Models and how Generative AI works (which I'm sure that most of you have by now), we provide a breakdown so that readers of different profiles can focus on their own areas of interest.
Recommended for everyone
The sections below provide definitions and set the baseline understanding of what Agentic AI is. Reading these will allow you to familiarise yourself with the language and terminology used throughout the rest of the document:
Readers who are interested in using Agentic AI
The following sections describe both existing and potential use cases, with a heavier focus on public sector ones. With reference to specific examples, you can better appreciate the strengths and limitations of Agentic AI technology. This will help you make a better judgement on whether your use case is suitable, or guide you in discovering suitable ones.
Readers who are interested in developing Agentic AI solutions
These sections focus on the "how" and provide examples of design patterns, frameworks, protocols, and platforms which can be used to develop Agentic AI solutions. The last section, Capability Development Areas, further discusses the steps we can take to close the gaps mentioned in Challenges in Using Agents in order to use Agentic AI technologies more effectively.
- Agentic Workflow Patterns
- Relationship with Other Techniques to Improve LLM-powered systems
- Implementation
- Capability Development Areas
Readers who are extremely busy
A TL;DR summary at the bottom of each section contains the key takeaways.