Mastering AI agents recap: Production-ready.
A hands-on, in-person workshop on building production-ready AI agents with open source software, covering Strands, Crew AI, LangGraph and MCP on Amazon Bedrock and Amazon SageMaker AI. Here is what I took from it.
The takeaway first, the tour after.
Three things I would tell a founder who could not make it.
I was proud to see that my clients and I had already built what they were showing, well before many big companies caught on.
The sessions didn't teach me much new technical content. The real value came from the connections I made.
The workshop confirmed we are ahead of the market in AI agent work. Many companies are still experimenting with basic LLM applications.
One day, start to finish.
Hosted at the Amazon office in Dallas. The event reached capacity and registration closed.
From building agents to cost and business value.
The workshop covered building agents with foundation models from Amazon Bedrock and Amazon SageMaker AI, and with open source frameworks including Strands, Crew AI, LangGraph and MCP Server.
The day included text to SQL agents, security analysis using Model Context Protocol in OpenSearch, evaluating agents, observability and tracing with Langfuse, and total cost of ownership and business value.
Couldn’t make it to Mastering AI Agents?
Get the recap and key takeaways in your inbox. Everything you missed, condensed into one resource.
- Event recap: key insights and takeaways
- Presentation slides (PDF)
- Links to tools and resources mentioned
This was an advanced hands-on workshop focused on building production-ready AI agents using open source frameworks including Strands, Crew AI, and LangGraph, along with large language models in Amazon Bedrock and Amazon SageMaker AI.
The workshop was held on Friday, June 20, 2025, from 8:30 AM to 5:30 PM CDT at the Amazon Office (DFW11) located at 13455 Noel Road, Suite 1400/14th Floor, Dallas, TX 75240.
The workshop was designed for developers, software engineers, development managers, data scientists, AI/ML engineers and architects, engineering managers and directors, and CTOs who are building GenAI applications and already have a basic understanding of agent concepts.
Topics included building agents using foundation models from Amazon Bedrock and SageMaker AI, using open source frameworks (Strands, Crew AI, LangGraph, MCP Server), text to SQL agents, security analysis using Model Context Protocol, observability with Langfuse, and calculating total cost of ownership and business value.
Girish observed that his clients had already implemented much of what was being demonstrated before many large companies. The real value came from networking connections with enterprise leaders beginning their AI agent journey and understanding how other companies are implementing AI.
Common challenges discussed included hallucinations, data security concerns, and cost management. Girish's experience addressing these issues through robust evaluation frameworks, secure data handling, and efficient prompt engineering positions him well to help others overcome these challenges.
No, this event reached capacity and registration is now closed. However, you can schedule a consultation with Girish to learn about production-ready AI agent implementation for your business.
You can schedule a consultation at gkotte.com/calendar to discuss how AI agents can transform your operations. Girish has already deployed multi-agent systems that handle complex workflows for clients and can help guide your implementation journey.
Put agents into production.
Focused use cases, evaluation and cost control. Bring the idea; leave with the next three steps.