This is the first blog post here. Let’s see what’s going on.
My name is Serge Yakupov. I live in Portugal with my family and all of my life I’ve been working in news media industry — since 2002. So news media is already in my DNA. I do startup at the intersection of media and AI — BotsEver. Actually this startup made me start learning Python. And this led me to Newslettrs.
I’ve started with Python In February 2023. The idea was just to do something that will help me in my everyday professional and personal life. And then I started to think about what this could be. I do read a lot of news about media industry. And no news aggregator I’m using 100% meets my requirements. Also I do really like newsletters. So the idea to combine these two things seemed to be obvious. Our mailboxes are still the best news aggregator we have.
I came to ChatGPT (well, yes) and asked: write python code to create rss reader. Obviously the response was very poor. And by the way during all that time I was creating Newslettrs, ChatGPT was really not 100% okay in coding. Every snippet had a mistake, outdated solutions etc. So I’ve learned a lot while correcting those mistakes.
Then I created a plan, a roadmap — what I wanted to have at the end. And one of the most important thing during creating the project, was — training the skill to write effective prompts. In general I’ve written more than 800 prompts to complete the code.
Long story short. It took a month to create some kind of framework where I can change parameters and get news, podcasts, newsletters on pretty much any topics. For a start I’ve chosen news about media industry and AI at first — the topics I’m interested in most of all. Than I did two more niche newsletters — about climate changes and about good news (we all need this kind of news, right?).
Then newslettrs.media appeared. I wrote some code to connect email engine and Stripe (the newsletters are free for the first month and then subscribers will decide how much it worth paying). I established a web server to run scripts according to the timetable I need.
And finally I thought that this could be some kind of side hustle for me (and maybe — my second startup).
What is Newslettrs now:
- It collects news on the topics (using keywords and list of sources).
- It creates smart summary of every news.
- It adds previews for new podcasts, blogs and newsletters.
- It packs everything in the letter for you to have all in one place.
- It sends emails.
So for now — Newslettrs is the bunch of fully automated newsletters on different topics.
Under the hood
It is Python with about 1000 lines of code. I asked LLM to read all news, search for the particular news I need (keywords, topics) and create summary of every news. Initially I wanted to use RSS feeds, but first — not every news source has this kind of feed and second — not all feeds are properly configured. Apart of news, Newslettrs collects relevant podcasts, blogs and newsletters to cover some expertise not just hard news.
Also we have about a hundred resources on different topics.
The roadmap for Newslettrs looks like this:
- Polish the engine to have better and accurate automatically curated news feed using machine learning.
- Add more niche newsletters — about esport, about other industries and topics (also about music and weekly recipes and cocktails recommendations for the paid subscribers).
- Add more tools and features to every newsletter we produce.
- Develop the project as a service with custom newsletters that everyone can create for yourself.
But the most important (and interesting) thing is to create an additional service that will let users to create curated automated newsletters to send them from their own name. That is my goal and main idea of the Newslettrs.
If you would like to talk about it — feel free to visit my LinkedIn page or drop me an email on firstname.lastname@example.org.
And subscribe if you think some of Newslettrs could be useful for you.