Swetugg 2026: Navigating the Edge of AI, Infrastructure, and .NET 10

Fresh off the energy of Swetugg 2026 in Liljeholmen, it’s clear that while the AI hype train is still roaring, the community's focus is shifting toward pragmatic implementation, sustainable infrastructure, and the robust new capabilities of .NET 10.

From the depths of the ocean to the cutting edge of local AI, here are the technical highlights and architectural takeaways from two days of deep dives.

Day 1: Modern APIs, Distributed Systems, and Edge AI

Mastering OpenAPI in .NET 10

Speaker: Sander ten Brinke

As we move into the .NET 10 era, the way we document and consume APIs is becoming more integrated into the development lifecycle.

Sander emphasized moving beyond basic documentation toward a "build-time" approach. By using the Microsoft.AspNetCore.OpenApi package (and the ApiDescription.Server NuGet), developers can now generate API documentation during the build process.

This is a gamechanger for PR reviews, allowing teams to see exactly how an endpoint's contract has changed before it ever hits a staging environment.

Technical Deep Dive: TypedResults vs. ProducesResponseType

• TypedResults: In .NET 10, favor TypedResults. It provides strong typing for your responses, making your code cleaner and allowing the OpenAPI generator to infer response types automatically without extra attributes.

• ProducesResponseType: Keep this for legacy controllers where you need to explicitly define metadata for the OpenAPI document, though it’s increasingly seen as the more verbose, manual way.

Tool Spotlight: Developer Flow

• CodeTour: Use this VS Code Extension to create guided walkthroughs of your codebase.

• DemoTime: Perfect for presenters, this tool helps automate VS Code setups for live demos.

Resources & Demos:

OpenAPI Tools Curator

ASP.NET Core OpenAPI Docs

Demo: OpenAPI .NET Talk GitHub

Distributed Systems with Aspire and Dapr

Speaker: Nico Vermeir

Nico explored the "Trenches" of distributed architecture, focusing on the synergy between .NET Aspire (the orchestrator) and Dapr (the runtime). The core takeaway: Aspire simplifies the inner-loop development, while Dapr handles the "sidecar" complexities of state management and service invocation.

• Dapr Extension for VS Code: An essential for debugging sidecar communication.

• Container Apps: The natural landing spot for these workloads in production.

Resources & Demos:

Aspire Dapr Demo GitHub

Official Aspire Docs

Offline AI in .NET MAUI

Speaker: Codrina Merigo

AI doesn't always need the cloud. Codrina showcased how to build privacy-first, offline-capable AI experiences using .NET MAUI. The focus was on local execution via ONNX and tools like CustomVision.com for model training.

• Netron.app: An indispensable visualizer for neural network models.

• Pocket Pal AI: A great reference for local LLM implementation.

Extra Learning: Demoing Mobile

Use Vysor to mirror and demo your local phone screen during presentations, it’s significantly more reliable than standard emulators for AI workloads.

The Undersea Cable Network: The Internet's Physical Backbone

Speaker: Richard Campbell

Watch the Full Session

Richard Campbell took us on a fascinating "Geekout" into the literal physical connections that hold the world together. Despite our obsession with satellites, 97-98% of all global data travels through over 2 million kilometers of undersea cables.

The Power Paradox: Modern cables aren't just for data; they carry high-voltage electricity to power repeaters every 50-100km to boost the signal as light scatters inside the glass.

The "Accidental" Weaponization: Campbell detailed recent incidents in the Baltic and Red Seas where anchor drags, some suspiciously timed, have caused tens of millions in damages, proving how vulnerable our "invisible" infrastructure truly is.

Day 2: Algorithms, Reusable Code, and the AI Reality Check

Algorithms Demystified

Speaker: Dylan Beattie

Dylan reminded us that "Algorithms are just spells", sequences of instructions that transform reality. From the classic Dijkstra's Algorithm (playfully explained as a Pub Crawl) to the philosophy of Zen and the Art of Motorcycle Maintenance, the session was a call to return to the fundamentals of logic.

Dylan’s Talk Resources

Building World-Class NuGet Packages

Speaker: Dennis Doomen

Dennis provided a masterclass in library maintenance. If you're building reusable code, your Directory.build.props should be your best friend to ensure consistency across projects.

The "Library Starter Kit" Checklist:

• PublicApiGenerator: To track and prevent accidental breaking changes in your public contract.

• Stryker Mutation Testing: Because 100% code coverage doesn't mean your tests are effective.

• Poly#: Use this to leverage modern C# features in libraries that need to support older target frameworks.

Resources & Demos:

NET Library Starter Kit

AI Hype: What’s Real and What’s Next

Speaker: Richard Campbell

Watch the Full Session

In his second session, Richard cut through the marketing. The term "Artificial Intelligence" was a marketing term coined in 1955 to raise military funding, and we are currently riding a massive "Hype Cycle".

• The Scaling Law: We’ve moved from the "bigger is better" mindset (trillion-parameter models) to a focus on smaller, specialized models that reduce "hallucinations" by narrowing the scope of data.

• The Developer Advantage: Software engineers are uniquely positioned to succeed with AI because we are already comfortable with test-driven development and code reviews, the perfect framework for validating AI-generated output.

Final Thoughts

Swetugg 2026 was a reminder that while the tools change (from Lisp in the 70s to .NET 10 today), the core mission remains: building reliable, sustainable systems. Whether it’s burying cables 4 meters deep to avoid anchors or using Aspire to manage your microservices, the focus is squarely on resilience.