Hi 👋🏼, I am Niklas, a DevOps engineer from Cologne, Germany.

Over the last few days I have been building a tool that suggests possible titles, meta descriptions, a LinkedIn post and a newsletter issue based on my blog post files in Markdown. Personally, I want to focus on writing good blog posts that reflect my thoughts or work. As the content is more important to me than the 'marketing' around it, be it the LinkedIn post or the 'perfect' description of the article, I have considered using the tool to have this suggested by an AI model. My blog post is always used as the foundation, which is always attached to the prompt.

As a result, I get suggestions from the AI which, in the best case (which is rare), I like straight away. In most cases, these are initial drafts, which I then modify as necessary to make them more suitable for my particular formulation. (Even if I prompt the AI to write them in the style of the blog post) - I like these suggestions as food for thought to add points to my post if applicable.

I see the AI in this case as a sparring partner to get the best solution for the task at hand. As a result, I don't spend as much time on the title, LinkedIn post, etc. as I do on the blog post itself. I tend to focus more on the quality of the blog post as the tool's suggestions are based on that.

It's similar with newsletters. I try to write it without improvement tools (DeepL Write or similar), but rather in the style I think it should be. The AI suggestions help me to find a rough structure for the newsletter on the topic.

I think this little app shows how we can use AI to 'simplify' our lives and I can focus on topics that are more interesting to me than writing the 'perfect LinkedIn post'. The quality of the content on my own website is more important to me personally than that on external platforms. In this case, AI helps me rather than replacing me completely.

I am currently using the Gemini 2.0 experimental model as a model for the tool - as it is free to use. However, I would like to test and compare Claude's results soon.

Update: I forgot to add an screenshot of the tool in action. Here it is:

Thanks for reading!

Niklas

When creating apps and websites, we often need to display data in different languages. Be it currencies, speeds, times or dates. Sorting words in different languages or dynamically setting the singular and plural forms of a word are also such use cases. For this there is the Intl API in JavaScript, which in my opinion still gets far too little attention.

In the past, I have written custom functions or used libraries to do the above in different languages.

So let's take a look at some of the functions of the Intl API and try to understand why it can make our lives easier.

Working with date objects in Javascript can be quite a pain. That is why I usually use a library like date-fns for comparing, adding hours to my dates or working with dates and times in general.

The TC39, which is the group of people who develop the definition of JavaScript, is working on the Temporal API. You can check out the proposal and documentation here.

With this, we no longer need date libraries, because the API itself is a breeze to work with.

I recently tried PGlite because I saw it trending on Hacker News. The idea is cool, having a built-in Postgres database that can even run in the client's browser brings a lot of possibilities. Seeing WASM (Web Assembly) get more use cases is also exciting to see.

Hey there, 👋

I needed to set up a Datadog agent to check remote Postgres instances as a regular container. All the documentation I found was about injecting secrets like passwords or hostnames via Kubernetes secrets. This seemed to be the only way to do it besides using Docker Swarm secrets. Since I had to set up a container definition for AWS ECS Fargate, neither of these options were available. So I wanted to see if I could use "regular" environment variables from which the agent could pull the secrets on startup.