Scotrix

I run 3 servers for 200 EUR, thanks to Hetzner, exactly for this reason (and I’m cheap and I never understood cloud/services like Vercel and Railway as serious alternatives ;-)).

Chuck Norris doesn’t die.

Quite a beast for a very good price…

I played a lot with LLMs over the last year and built a multitude of products with it and it’s always just the same bottom line outcome: - be specific - keep it small - be precise when adding context - don’t expect magic and shift deterministic requirements to deterministic code execution layers

All this is awfully painful to manage with current frameworks and SDKs, somehow a weird mix of over-engineered stuff while missing the actual point of making things traceable and easy changeable once it gets complex (my unpopular personal opinion, sorry). So I have built something out of my own need and started to offer it (quite successfully so far)to family & friends to get a handle on it: Have a look: https://llm-flow-designer.com

While working on a new startup I had exactly the same challenges and issues. As soon as it gets complex, data gets bigger, amount of tools increases it becomes tremendously difficult to control agents, network of agents or anything LLM related. Add then specific domains like legal or finance where most of the time there is just 0 or 1 and nothing in between (metaphorical speaking) it becomes a nightmare in code and side effects.

So I started to actually build something to solve most of my own problems reliably and pushing deterministic outputs with help of LLMs (e.g. imagine finding the right columns/sheets in massive spreadsheets to create tool execution flows and fine tuning finding a range of data sources). My idea and solution which helped not only me but also quite a few business folks so far to fix and test agents is visualizing flows, connect and extract data visually, test and deploy changes in real time while keeping it very close to static types and predictable output (other than e.g. llama flow).

Would love to hear your thoughts about my approach: https://llm-flow-designer.com

Just find a Hoster with low traffic egress cost, reverse proxy normal traffic to Cloudflare and reply with 2GB files for the bot, they annoy you/cost you money, make them pay.

would be good to have an indicator if it’s available with your distro by default or what package you’ll need to install it since all tools are only as useful as available they are…

+1 I dropped school relatively early (I was extremely bored and the way of education was certainly somewhere close from the Stone Age times). I did an apprenticeship as software engineer with some (extremely useless) school component. Most of the time in the late 90ties was trial and error, for me, the master and the master of masters. Playing around with Linux and make it ISDN routers with servers for websites built in HTML, Perl, PHP. This was devops (before it got hyped) and real engineering by figuring stuff out with almost no documentation, a lot of crazy creativity and push the boundaries of what’s possible. And it reminds me just a little like today’s world with AI and vibe coding just on a complete different level and with significant more pressure…fun times :-).

cool, i'll have a look. thanks for sharing.

I input text and preferably I output JSON but doesn’t matter much as long as it’s somewhat structured.

Ultimately I’d like to extract information like date ranges, specific indications of tool usages (e.g. I have a bunch of data apis with their own individual data and semantic meaning which need to be picked and then a combination of tools to transform the data)

The first few seconds/minutes I felt the same but it was fine quite fast. I think the typical spotlight reindexing was the culprit.

I switched as well and I actually use the AI assistant since then primarily. It’s awesome to connect search directly with AI, almost always get what I want immediately.

yeah, but it doesn’t need to be that way. It can be simple and makes it easier adoptable, why over engineering and reinventing the wheel of at least 2 decades experience and better practices.

OpenAI takes the backseat and wait until something stable/usable comes out of it which gains traction and takes it over then. Old classic playbook to let others make the mistakes and profit from it…

Couldn’t agree more, played the whole day today trying to get a HTTP MCP server with Claude running.

Absolutely terrible, no clear spec, absolute useless errors and/or just broken behaviour without telling what’s wrong. Reference implementations and frameworks are not working either, so only reverse engineering + trial & error until it runs, yaaay.

Feels like the early 2000 over and over again, trying to make something work.

wasn’t vo1d something which could potentially deliver in this?

well, min. required network confirmations could still replace the progress bar. I certainly pressed reload anxiously quite a few times :-)

No auth smtp server sounds like a very bad idea and the real culprit here. Security by obscurity (VPN in this case) never works.

This story is for me a real painful death I have experienced way too many times, absolutely nuts.

But then you find an open source solution which is in general better and can do everything you want (simply tested already with just a docker compose up) but for deployment you get hit hard by compliance who just checks the SOC2 certifications and wants a in-depth due diligence of the code since everyone in the world can theoretically change it. Then your manager asks how it can be so good if it's for free and open source. And of course, last but not least, your overloaded team in general not happy to support just another unpredictable piece of software...

So it's the question to rather burn money and nerves with an awful SaaS offering and their endless and useless sales cycles and terrible and super expensive vendor-lock-ins or burn some money and nerves by utilising and running open source inhouse...

So typically I prefer to chose for the open source option and especially if the SaaS option isn't allowing me easy and fast self-onboarding, meaningful testing periods and a predictable and transparent pricing.

And then, if it get's widely adopted, I allocate some budget to support the authors and/or get some support plan (for more complex open source software) in place even though you most likely never need it...

Agreed, cool and certainly some improvement but Freestyle is good enough already. Next step for me is more towards the new insulin research which activates only on glucose in the bloodstream (don’t recall how it was called but was here more recently also shared). This sounded more self controlling where it’s hopefully just making sure you have enough of it in your body and don’t need to take care of the rest.

Until then, Freestyle with Omnipod Dash in a close loop with iAPS was a game changer for me: Almost no peaks anymore, HBA1c on the level of a non diabetic person…

Nevertheless, good luck in productising it and I’ll be certainly trying it once it’s available…

“Asking our governments to create laws to protect us is much easier than…”

A bit naive that, it’s too late since data is already mostly available and it just takes a different government to make this protection obsolete.

That’s why we Germans/Europeans have tried to fight data collections and for protections for so long and quite hard (and probably have one of the most sophisticated policies and regulations in place) but over time it just becomes an impossibility to keep data collections as low as possible (first small exceptions for in itself very valid reasons, then more and more participants and normalization until there is no protection left…)

Very nicely written article. I’m also running a k8s cluster but on bare metal and qemu-kvms for the base load. Wonder why you would chose VMs instead of bare metal if you looking for cost optimisation (additional overhead maybe?), could you share more about this or did I miss it?

Regulation, Taxes, inflexibility and small-thinking mindset let me leave Germany 10 years ago. I didn’t regret it a single day and everytime I’m going back home after visiting my parents in Germany I feel so much more free.

very nice!

I built a service (https://yasl.at) which allows to generate short url with customised meta data via a simple API to deep-link into web games and specifically results/things worthwhile to share without having the headache of tampering in each game over and over again.

You can read more about one of my use-cases specifically here: https://medium.com/@degola/supercharge-your-links-with-custo...

Open to do some backlinking @fayazara?

As a developer I’m always surprised that these issues are real.

I’m used to store raw data I receive back from payment processing in my own database as transactional events (e.g. renew, cancel, successful and failed payments ) ends up as single events. I do the same for all application events and it is then rather easy for me to get the data as I need it for whatever KPI I want (activity, retention, usage) and for very little overhead while building products.

So I’m wondering, is this a result of all the low-/no-code stuff going on?

I’m working on a few things:

- Micro SaaS https://yasl.at, link shortening with custom meta data for link previews on messengers/social media, see my blog post for more details: https://yasl.at/dKpNnZ

- A web game called Wizencraft (https://wizencraft.com) which is a logic puzzle to guess the right order of items as quickly as possible which lets you compete with other players, solve daily challenges in groups and let’s you grind thousands of items, it's a very early alpha and I'm building it in public very iteratively and not polished at first, see more details here: https://yasl.at/XnoDtY

- And last but not least I started to build a not yet published simple dashboard for analytics where I can plugin my self-hosted database very easily. You can just write SQL queries for MongoDB, Clickhouse, Postgresql in a single YAML which defines the query and how a chart looks like on the dashboard.

The idea came from the fact that using other dashboard/analytic tools always is a hassle, existing analytics dashboards I used for products are not easy to copy, replicate or adjust and it's mostly doable solely via clicking in a complex frontend instead of defining it simply via a configuration and the available offerings are charging large.

I'm planning to have the chart definition editable via web interface and is stored locally as files which can be versioned and managed by git and can connected to the web interface with a simple docker container (this brings the analytics dashboard to the product source code itself and can be easily copied/adjusted again for new products). Happy to share more and once the first version of the dashboard is ready to use (in the coming weeks...).

---

Would love your thoughts on all of this and especially the analytics dashboard. Did you you face similar issues/problems with existing solutions (or if i can stop and actually use an existing solution which does what I'd like to have).

I‘m T1D and using Freestyle Libre + Omnipod Dash and iAPS + Apple Watch. Apple Watch is for me primarily to automate physical exercise detection and target adjustments but also works great with iAPS to control bgs and inject insulin from your watch without taking your phone out of pocket. All built as a homebrew closed loop.

While it was somewhat difficult initially to make it work I managed to get over the last year to 85% in range continuously over weeks with a (for me in comparison to before) very low amount of hypos (3 or 4 per week).

Happy to share more and the challenges I had if someone is interested...