Skip to content

Blog

AI Agents + British Ryan

AI Agents + British Ryan
4:31

 

Dr. Ryan Ries here, and this week, I want to dive into a topic I’ve been seeing pop up a lot recently: AI agents.

AI agents were the talk of the town at both HubSpot's Inbound conference and Salesforce's Dreamforce, where both companies unveiled new AI agents in their products. We have also seen large model companies and cloud providers adding agents into all of their frameworks. But what exactly are AI agents, and why should you care? 

What Are AI Agents?

AI agents are specialized AI systems designed to perform specific tasks or functions autonomously. If you were big on Robotic Process Automation (RPA), then this will give you that Deja Vu feeling.

Think of them as digital employees, each with their own job description and area of expertise.

Unlike chatbots or simple AI assistants, these agents can make decisions, take actions, and even learn from their experiences.

AI Agents vs. Custom GPTs vs. Chatbots: What's the Difference?

While custom GPTs are tailored language models designed for specific conversational tasks, AI agents take it a step further. They can:

  • Interact with various systems and databases
  • Make decisions based on complex criteria
  • Perform actions without constant human oversight
  • Learn and improve their performance over time

Essentially, if a custom GPT is a smart conversation partner, an AI agent is more like a virtual coworker.

Another way to look at it: An AI agent is ACTIVE, while a chatbot and custom GPT are PASSIVE.

Amazon's Take on AI Agents

Amazon's Bedrock Agents are pushing the envelope even further. We love the Amazon Bedrock Agents and use them all the time. Here are some of their features:

  • Retrieval Augmented Generation (RAG): These agents can securely connect to your company's data sources, pulling in relevant information to generate accurate responses.
  • Multistep Task Orchestration: Bedrock Agents can break down complex tasks into logical sequences, calling necessary APIs to interact with company systems and processes.
  • Code Interpretation: They can dynamically generate and execute code in a secure environment, enabling complex analytical queries and data visualizations.
  • Memory Retention: Bedrock Agents can remember historical interactions, providing more personalized and seamless user experiences.
  • Chain-of-Thought Reasoning: You can trace through the agent's reasoning process, helping you troubleshoot and improve its performance.

The Future of AI Agents: My 2025 Predictions (and Beyond)

As we look ahead to 2025, here are some of my predictions for AI agents:

Prompt Configuration Management

Right now, everyone is trying to figure out how to store, maintain, and update their prompts.

People are often just putting them into Python notebooks and storing them in GIT repositories, but it doesn’t feel quite right.

My team has been working through methodologies and trying to figure out the best way to track and update prompts.

This will be a big topic going forward, and we will see new systems coming online to help.

Autonomous Workflow Management

Autonomous workflow management refers to AI agents taking on entire business processes, from initial customer contact to final delivery, with minimal human intervention.

I don’t think this is fully a 2025 thing — but we can expect to see autonomous workflow management crop up more and more in the next 5 years.

New Job Opportunities

Yes, AI agents will start to work autonomously.

As previously mentioned, they will eventually need minimal human intervention.

But, this means we will need a new kind of role: AI agent managers.

Someone who will make sure these AI agents perform their tasks and is able to fine-tune these agents over time.

What Do You Think?

Are you excited about the potential of AI agents, or do you have concerns? How do you see them fitting into your industry or workplace? Do you have any other predictions for the next year or so? I'm curious to hear your thoughts.

Until next time,

Ryan Ries

Now, for this week’s AI-generated image, I wanted to do something a little different!

I’ve been hearing Runway AI pop up a lot lately as well, so I figured I’d take it for a spin. On the free tier, you can’t use your own voice, so here’s me with a British accent.

 

 

 

Sign up for Ryan's weekly newsletter to get early access to his content every Wednesday.

Author Spotlight:

Ryan Ries

Keep Up To Date With AWS News

Stay up to date with the latest AWS services, latest architecture, cloud-native solutions and more.