Martin Adamek

Software engineer running Porsche service and restoration shop. And buildings apps for it.

[Martin Adamek @stuartisan]

@aditya520e @openclaw Yes, we are testing right now.

[Martin Adamek @stuartisan]

Did anyone manage to tame @openclaw in multi-user environment? Agent per user. Separate user and team resources. Multiple teams on single machine. Testing such setup right now, about to deploy for the company as well as family.

[Martin Adamek @stuartisan]

My son made a model of my 911. Lot of work and skill in Blender.

[Martin Adamek @stuartisan]

Gemini 2.5 Flash surprisingly good at workshop manual parsing. Mixed orientations, illustrations, photos and yet the Markdown version is perfect. Good source for QMD and company’s OpenClaw.

[Martin Adamek @stuartisan]

Hetzner VPS + tmux + Termius + Tailscale is my dev machine. No local dev environment anymore.

[Martin Adamek @stuartisan]

@TermiusHQ Can we get the notification after agent finishes work instead of every text new output? 🙏

[Martin Adamek @stuartisan]

I really to iterate designs with Claude's /frontend-design. I ask it to shoot 3 static designs at /ideas page and go from there.

[Martin Adamek @stuartisan]

Working on several apps in parallel created a new need for me - separate workspace for each app. I need a named display with terminal + browser. Would be much easier on Linux, but on Mac it is the AeroSpace by @nikitabobko that works for me.
https://t.co/rcX0ulQ0VN

[Martin Adamek @stuartisan]

Having Rails dev environment locally is so annoying now that I tasted Tailscale + Tmux + Termius combo.

[Martin Adamek @stuartisan]

@schwad_rb @hatchboxio Same!

[Martin Adamek @stuartisan]

Nice, I finally configured this inline diff in LazyGit (and LazyVim). Much better. More important now with agent coding. Also standing desks.

[Martin Adamek @stuartisan]

@quatermain32 Interestingly, my new (much simpler prompt) takes a bit longer, but gets perfect response. Or maybe I need to run it at least 10 times to avoid some random delays. Whatever, it is faster and 100% covering my test, so good on my side ;-)

[Martin Adamek @stuartisan]

Impressed by Haiku 4.5 - invoice parsing times down to half comparing to 3.5 and parsed all the test files correctly. Currently I can't find any invoice, that would not be 100% perfectly extracted. Also, simplified the prompt by [+15 -73].

[Martin Adamek @stuartisan]

@quatermain32 Claude Code did that, not me ;-)

[Martin Adamek @stuartisan]

@quatermain32

[Martin Adamek @stuartisan]

@quatermain32 Just simple prompt and JSON output.

[Martin Adamek @stuartisan]

@c_sonnier Take it with a grain of salt. I might have been to quick to conclusion.

[Martin Adamek @stuartisan]

Passing blob.path vs. temp_file.path is all the difference in perfect LLM invoice parsing. Using gemini-2.5-flash via ruby_llm. Is ActiveRecord messing the pdf somehow?

[Martin Adamek @stuartisan]

@c_sonnier Sure, just can’t DM you.

[Martin Adamek @stuartisan]

Today's progress on tuning of LLM-based PDF invoice parsing. Now I can add models and sample files at any time and feel a little bit under control. Thanks ruby_llm for making model switching efortless.

[Martin Adamek @stuartisan]

@mtantawy Tuning the prompt and comparing the results agains some golden files, to get as consistent results as possible.