Thoughts
This year's hacks and tricks from Swetugg Conference
Esatto's René Voigt went to Liljeholmen for this year's Swetugg Conference and returned with a wealth of trends, tips, tricks, and links that he shares here.
My main takeaways from the Swetugg Conference in Stockholm 2024
Disclaimer: I have not yet tested all of the mentioned technologies myself, but I hopefully will have time for that soon ;)
This year's hype and buzzwords
- Bicep: Infrastructure as code
- Gen AI: of course
- Having the coolest terminal theme
Better code reviews
https://mtlynch.io/human-code-reviews-1/
https://mtlynch.io/human-code-reviews-2/
https://mtlynch.io/code-review-love/
.Net Aspire (preview)
New framework from Microsoft to make orchestration of cloud-native apps simpler.
https://learn.microsoft.com/en-us/dotnet/aspire/get-started/aspire-overview
https://github.com/dotnet/aspire
Example
https://www.youtube.com/live/kAF9No5KZrg?si=YpZ_FdxA4I1QxFBX&t=1170
Related
Azure Container Apps: https://learn.microsoft.com/en-us/azure/container-apps/overview
Testing
Playwright
Microsoft's browser testing tool, similar to Selenium and Cypress
https://github.com/microsoft/playwright
Playwright Testing service
Running Playwright tests at scale
https://azure.microsoft.com/en-us/products/playwright-testing/
Azure testing
https://azure.microsoft.com/en-us/products/load-testing
https://azure.microsoft.com/en-us/products/chaos-studio
Accessibility
Microsoft Accessibility Insights: https://accessibilityinsights.io/
Microsoft Clarity
Free session recording and heatmap tool, similar to Hotjar
https://clarity.microsoft.com/
Related
Bicep linting: https://dev.to/lakkimartin/adding-linting-validation-and-what-if-to-bicep-templates-in-azure-devops-2cjp
PSRule for Azure: https://azure.github.io/PSRule.Rules.Azure/about/
Gen AI
Prompt context compression
Compress the prompt history to minimize the amount of tokens sent on each consecutive request. I could not find a similar approach online yet. Ask OpenAi to compress the text into a format that does not need to be human-readable.
Embedding models
OpenAi has released new embedding models that are both cheaper and more powerful.
https://openai.com/blog/new-embedding-models-and-api-updates
Bribe the AI
You can actually "bribe" the AI to get better results, examples here:
https://medium.com/@jordan_gibbs/a-comprehensive-guide-to-bribing-chatgpt-cfbefefc49a0 (partially pay-walled)
Azure Assistants
Create tailored AI assistants directly in Azure or by using the API.
https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/assistants
Open AI cookbook
Example code and guides for using the OpenAI API, the examples also contain lots of notebooks (see below)
https://github.com/openai/openai-cookbook
Other stuff
I will never get tired of ridiculous Apis:
VS Notebooks
Create or use interactive courses or why not document your API using Notebooks?
Azure durable functions
Create stateful, chainable and long-running functions