Each year, the Kansas City Developer Conference (KCDC) brings together software developers, technology leaders, and industry professionals to share ideas, explore emerging technologies, and learn from one another. KCDC 2026 was no exception. With dozens of sessions covering topics ranging from AI and security to cloud technologies, development practices, and modern application architecture, attendees were challenged to choose between an abundance of valuable learning opportunities.
Several members of the Omnitech team attended this year's conference and returned with new perspectives, practical ideas, and inspiration to apply in their daily work. While their session choices varied, a common theme emerged: the rapid evolution of AI and the growing opportunities it presents for developers, organizations, and the future of software development.
Below, Aaron, Chase, and Warren share their highlights, key takeaways, and reflections from KCDC 2026.
Attending the Kansas City Developers Conference (KCDC) was a great opportunity to take a break from familiar day-to-day work and spend a couple days learning about a wide variety of technologies and ideas. The amount of content available was almost overwhelming... in a good way! There were so many interesting sessions in every time slot that one of the hardest parts of the conference was deciding which ones I was willing to miss.
A few sessions especially stood out to me. “Correcting Common Async/Await Mistakes in .NET” was a great deep dive into something I use regularly but can always understand better. I also really enjoyed “The Missing Piece in Agentic AI Security,” which focused on fine-grained authorization and how applications can control exactly what AI agents are allowed to access and do. “Modern Auth Patterns in .NET” covered a number of current authentication and authorization practices that are increasingly important as application architectures evolve.
Another favorite was “Just Use Postgres,” which explored the extent of PostgreSQL's functionality beyond that of traditional relational databases. Coming primarily from a SQL Server background, I found it interesting that PostgreSQL can also handle document data, full-text and vector search, geospatial data, job queues, and other workloads that might otherwise require additional services.
More than anything, conferences like KCDC are a good reminder of how much there is still to learn and how many different approaches there are to solving the problems we run into as developers. I came home with plenty of ideas and topics I want to explore further… and, of course, any trip to Kansas City that offers a chance to stop at Q39 for barbecue is always welcome.
KCDC 2026 was a great couple of days to step away from my normal work and see what other developers are building and thinking about. The schedule was packed. I bounced between web components, web workers, Cloudflare Workers, Postgres, and a lot of AI talks, and I came home with a notes folder that's a lot fuller than I expected.
My favorite session was "Old API, New Tricks," which covered adding MCP (Model Context Protocol) to existing .NET REST endpoints. Put simply, MCP is an open protocol that lets AI agents talk to outside systems. The speaker compared it to the REST APIs we already built. Instead of reading OpenAPI docs, an agent discovers tools at runtime. Instead of HTTP verbs and status codes, it calls named tools that return JSON schema results. What I liked most is how practical it felt. The C# SDK lets you mark up existing classes and methods with attributes so they can be discovered, which means you don't have to rewrite your API to make it AI-friendly. One line that stuck with me was "that a tool's description is your contract with the agent. If the description is vague, the agent will be too" (which feels like good AI advice in general). He also covered the recent security improvements, like OAuth 2.1 support and audience-bound tokens. The audience of this talk also gave me insight into just how many different developers/companies were using MCP.
My other favorite was "AI Sandboxes," which was about running coding agents safely. Once you give an agent access to your files, tools, and credentials, you've opened up a lot of ways for things to go wrong. The talk broke isolation into four areas: filesystem, network, credentials, and processes. The speaker also made a fair point that too much isolation creates so much friction that people just stop using it. Docker Sandboxes handle this by running the agent in a microVM, with every outgoing request routed through a host proxy that either allows or blocks it. My favorite detail was that credentials are never read directly from the host, so even a compromised agent can't read its own token.
Overall, KCDC gave me plenty of ideas and tools to continue diving deeper and implementing them in my projects. It was also a good reminder of how fast things are moving and how important continuous learning is, especially around AI.
I had a great time at KCDC 2026. It was my first time attending this conference, and it lived up to its reputation as the premier engineering conference in the Midwest.
Something that stood out to me was seeing how extensively other professionals utilized AI in their development workflows. Many of the sessions I attended touched on the use of AI in one way or another.
The "AI Sandboxes" session focused on the secure use of AI agents on your machine. The idea is to use Docker Sandboxes to lock down the AI agent to a specific directory and limit access to external resources.
One of my favorite talks was about using AI for end-to-end testing, which involved creating and maintaining a "Playbook" that the AI agent would use to generate tests. I left inspired to explore this area further for my everyday work.
Another interesting session was on Spec. Driven Development - the developer writes specifications for a piece of software, AI builds the software implementation, and the developer reviews the result and refines the specifications accordingly. The speaker described working on a project where he hadn't looked at a single line of generated code.
I am excited to start implementing some of the tools and ideas I learned about at KCDC.
