What the heck is an AI agent?
👋 Welcome to New Vintage, a weekly taste of tech for wine professionals to level up with no-code and AI tools.
If you enjoy this newsletter, please forward it to a colleague – every new reader means so much!
AI AGENTS are the hottest trend in tech and they are here to stay.
Imagine having a friendly, ever-patient, co-worker that helps you with the jobs you find boring or time-consuming. That’s pretty much what an AI agent is!
It defines tasks based on the goals you give it, gathers the information it needs, and takes action—even checking its own work as it goes.
Let’s explore what agents are, how they work, and a simple way to dip your toe in without diving right into advanced AI.
Why This Matters Now
Simple Reason: Wine brands are often understaffed and AI agents are new systems that will do more and more "human" work.
Maybe you’ve seen news headlines about AI taking over jobs or doing things faster than humans ever could. AI will definitely change the way we work and there will be growing pains.
While that might sound scary or exciting, the idea behind AI agents is actually simple: the intention is to save you time. By handing over repetitive tasks to an AI agent, the goal is to free up more of our time for higher level work (or leisure time!).
This could be welcome relief at a time when it feels harder than ever to build a thriving wine brand.
So...What Is an AI Agent?
AI researchers and developers argue over the precise definition of an agent, but I think of it as a new type of system intended to be more like a digital co-worker than traditional software.
The typical distinction used is that AI Agents can act autonomously, where traditional software would be limited to what it was explicitly coded to do.
This can mean many things, so I like how Harrison Chase (CEO of Langchain) thinks about how "agentic" a system is to describe how much autonomy is intended.
Much like a co-worker, agents can be given different goals, data, system access, and instructions to get work done. We are at the earliest stages of this stuff, so expect them to make mistakes, need oversight, and require software engineers to implement them.
Unlike a co-worker, they don't: fatigue, get bored, or sue you for wrongful termination. So there's that.
If you're curious about the current thinking on the archetypes of agents, AWS has a decent breakdown here.
How AI Agents Work
This is kind of a trippy concept, but the basic mental model of how agents "work" is largely how we work.
When an AI agent has a job to do, it usually follows four main steps:
- Figure Out the Goal: You tell the agent what you want done (like sorting emails or checking stock levels). It then splits that goal into smaller tasks.
- Gather Information: It looks for the data it needs—maybe from your computer, the internet, or other agents.
- Do the Tasks: Step by step, it handles each small job. Once it finishes a task, it checks to see if it needs anything else before moving on.
- Evaluate & Improve: The agent asks, “Am I done? Did I do this right? How can I improve next time?” This helps it learn and adapt.
Of course, this is an extreme simplification, but it's an important step in broadening our understanding of what is possible today and going forward.
Getting Started Without Going Full AI
You don’t have to jump straight into building your own AI agent.
You can practice by streamlining just one simple task in your day-to-day work.
Here’s how:
- Pick One Repetitive Task: Maybe you get the same email questions every day, or you always enter the same info into a spreadsheet.
- Write Down the Steps: Who does what? Which software do you use? Where does the information come from?
- Try a No-Code Tool: Use something like Make.com (or another “no-code” tool) to automate part of that task. For example, it could send automated replies or update a file for you.
- Tweak & Let It Run: See how much time you save. If it’s helpful, keep adjusting until it runs smoothly on its own.
The accumulation of these reps will give your team incremental gains in productivity, while building the foundation you'll need for deeper use of AI tools/systems.
Final Thoughts
AI Agents aren’t magical creatures—but they are new kinds of systems that learn, plan, and do tasks with minimal help from you. If the idea still seems a bit futuristic, start by automating one small part of your workflow. By the time you’re ready for a full-fledged AI agent, you’ll already have a good sense of how to set goals, gather data, and let the technology do the heavy lifting.
If you want me to dive deeper on any subjects, always feel free to comment on any posts.
Some related posts you might be interested in:
4 ways to get better output from ChatGPT
A primer on what happened in AI in 2024
Hope this was helpful,
Stephen
Comments ()