Timo Dechau

Tbh. Business processes are build on hallucinations. So it might fit well.

Now we all have to learn to ask the right business questions 😄

Unfortunately self-service BI goes beyond data access. But it is a cool step forward.

Silent mode for some weeks and plenty of refactoring for the tracking plan generator. Major change, swapped Supabase for Convex and Sonnet 3.7 for 4.0. It generated this for a Miro use case:

That looks really good. What kind of stack are you using?

Puh - the new Propagandhi album is going deep... Really good, but completely different from what I thought it would be.

Don't ask. I saw something that someone build for himself and I liked the approach so much, that I wanted something similar.
Then it started of with some vibe coding until it got to a specific stage.

Good day - part 2. First successful deployment of the home-built course platform. Some more tests, and we can move the people from Circle to it.

Good day - I generated the first test tracking plan via the app. My usual test case: Miro. 30s, 6 entities, 38 events (maybe a bit too many), 45 properties. Quality - quite close to my manual designs. Next step - make the table look nicer.

That is mostly a problem of scale. Human produced content sometimes intends meaning. Sociopathic content aka gossip and friends are making people dumber since 10k. The scary thing is the scale. That is true.

No - what I meant it was said, that it makes people dumber. And it was said about so many other things (like reading comics as a kid or YouTube). Therefore I don't believe in making dumb anymore.

It be worth a test. I completely agree with you about strong personal thoughts or storylines. Not so much about who writes it out. But everyone picks different tools. And has different objectives.

Was this not also said about newspapers and most likely printing press. It’s an evolution. Will it change this - definitely. Will all be good, no. But there will be good stuff.

What do you recommend someone who hates writing sentences? I love to break down thoughts, outline them, fine tune the story. But I don’t like to write it out.

Incremental with SQLmesh - oh la la.

I have a list with 80s and 90s action movies. I won't go for vinyl - this will become too expensive. You can watch the Steph Curry docu on Apple+ to upgrade yourself 😅

I guess my version of midlife crisis is thinking about buying band shirts again - I know a mild version. Talking to Johan van de Werken (not if you are on BS) does not help.

Ok, I am pushing it now - create tracking plans with multiple models, synthetize and do cherry picking to get your perfect tracking plan. But beware, I will show warnings when create more than 20 events. This is an opinionated software.

It looks so innocent in that stage - I could keep it for simplicity like this - but there is this urge for individual UI that cannot be controlled.

Why Unit Economics Outlast Growth Metrics When capital is cheap, any chart can look heroic. When markets tighten, only businesses with positive unit economics breathe easy.

I really enjoyed Juliana's latest post about unit economics - julianajackson.substack.com/p/unit-econo...

Unit economics is the adult version of analytics.

Me too.

This pixel art was already worth working on a course withJuliana J.

We plan to launch a course with the monumental title: Strategic Analytics.

Join the waitlist here: lp.timodechau.com/strategic_an...

Something I always wanted to have in a tracking plan tool: auto-population of descriptions and examples.

Now I added it to the HipsterDataPlan. My current optimization metric is time to the v1 version of a tracking plan. This saves me 30-60m easily.

Ok, a year ago, I would have said a definite no to building a tracking plan/setup tool. But things have changed. I now have a way to generate really high-quality tracking plans with GenAI, so now I'll build the frontend for it. I'll keep you posted.

Thanks to vibe coding - I do a session on the tool every morning and one every evening - adding a new feature idea and testing it throughout the day. Fine-tune and properly remove it again. And it has a shape already I would not have most likely on a traditional path.

I am currently experimenting with building a task management that works perfectly just for me. I add typical task features like priority or sizes and remove them again. And maybe introduce them in a different shape. A lot of back and forth.

One thing that I can't get out of my head. What if most digital products are compromises made too early? You rush to product market fit, but is your product that good to make a difference in the long term? Did you start to bake in features you are not passionate about because they seem reasonable?

Me, too. Especially the pipelines.

@evidence.dev @hughess.bsky.social - Your Slack channel can only be joined with an evidence email address. I have some questions about chaining source sql files..