well architected iac analyzer - Gen AI tool analyzes architecture diagrams (PNG, JPEG..), convert them into CloudFormation or Terraform code, and provides recommendations for optimization.

🧠 USE CASE

Understanding AI Agents Workflow

Most of you may recall that we discussed the surge in AI engineering roles and saw the trend of implications in the DevOps, Kubernetes, and SRE roles.

And I promised you to bring at least one write up per week along with the usual Kubernetes, DevOps, and Cloud use cases we normally discuss.

Last week, we started the AI Agents series with the ABCs of AI Agents

I suggest reading through it for basics if you missed out.

So far, we understand that an AI agent not only analyzes data but also understands problems, makes decisions, and takes real actions to solve them.

It learns, adapts, and works independently or alongside humans to automate tasks.

I've created this illustration to simplify how an AI agent works.

Inspired from: The Rise and Potential of Large Language Model Based Agents White paper

Download a high resolution pdf of this diagram here for future reference.

An AI Agent works by following a structured workflow that mimics human perception, decision making, and action execution.

Here's the short breakdown:

1. Perception (Data Ingestion & Processing)

An AI agent receives inputs from different sources such as text, images, videos, or external systems.

It processes text using Natural Language Processing (NLP), which involves breaking down sentences into meaningful parts to understand context.

Images and videos are analyzed using Computer Vision, which detects patterns and extracts relevant details.

This step ensures the agent understands context and intent before taking action.

2. Brain (Memory, Knowledge & Decision Making)

The agent stores and retrieves knowledge using vector databases (e.g., FAISS, Pinecone) and structured storage (e.g., PostgreSQL, Redis).

It makes decisions using predefined rules or machine learning models (e.g., LLMs), which learn from past experiences.

Some agents use reasoning engines like symbolic AI or knowledge graphs to improve decisions over time.

3. Action (Execution & Automation)

Once a decision is made, the agent performs actions such as generating structured responses (e.g., JSON, YAML), triggering API calls, or automating workflows in external systems like Kubernetes, CI/CD pipelines, or cloud services.

In short, when you ask an AI agent to check your Kubernetes cluster upgrade plan, it retrieves the cluster state, analyzes node compatibility, deprecated APIs, control plane version, and workload impact, then applies decision logic to assess risks.

Based on this, it would give you a detailed upgrade plan or warn you to not upgrade due to underline reason.

Exciting days ahead, isn’t it?

Let's keep this AI learning going! Up next, we will discuss:

Types of AI Agents
Risks of AI Agents
Roadmap to AI Engineering

Much more to explore

AI Agent Engineer roles are slowly coming to eat up DevOps and many other roles.
This isn’t speculation, the data backs it.
CEOs and CTOs started asking: “How can AI agents optimize our systems?”
This top down push is unconventional but happening.
Let this sink in, AI… x.com/i/web/status/1…
— Govardhana Miriyala Kannaiah (@govardhana_mk)
3:11 PM • Jan 29, 2025

I run a DevOps and Cloud consulting agency and have helped 17+ businesses, including Stanford, Hearst Corporation, CloudTruth, and more.

What people say after working with me: Genuine testimonials

When your business needs my services, book a free 1:1 business consultation.

Looking to promote your company, product, service, or event to 35,000+ TechOps Professionals? Let's work together.