from Heidelberg, Germany
Opinions are my own
I built a pipeline: k2pdfopt reflow → text-layer EPUB → push to my Xteink X3. Text now uses ~90% of the screen instead of ~35%.
dakoller.net/blog/2026092...
I built a pipeline: k2pdfopt reflow → text-layer EPUB → push to my Xteink X3. Text now uses ~90% of the screen instead of ~35%.
dakoller.net/blog/2026092...
I built an n8n adapter that keeps the credential server-side, filters events, and feeds a public JSON endpoint to our static website.
dakoller.net/blog/2026091...
#n8n #API #Automation#konzertmeister
I built an n8n adapter that keeps the credential server-side, filters events, and feeds a public JSON endpoint to our static website.
dakoller.net/blog/2026091...
#n8n #API #Automation#konzertmeister
It’s about blocked work, durable state, checkpoints, and why mature agents need to stop cleanly instead of guessing ahead.
dakoller.net/blog/papercl...
#AI #Paperclip
It’s about blocked work, durable state, checkpoints, and why mature agents need to stop cleanly instead of guessing ahead.
dakoller.net/blog/papercl...
#AI #Paperclip
This one is about bounded agency: why capable agents still need to know which decisions belong to their role, and when to route work elsewhere.
dakoller.net/blog/papercl...
#Paperclip #AIAgents #BuildInPublic
This one is about bounded agency: why capable agents still need to know which decisions belong to their role, and when to route work elsewhere.
dakoller.net/blog/papercl...
#Paperclip #AIAgents #BuildInPublic
Markdown stays the source of truth. A small generator builds the book structure. Quarto outputs PDF, EPUB, & HTML.
dakoller.net/blog/2026082...
#Paperclip #Quarto #Markdown #Documentation #PDF #EPUB
Markdown stays the source of truth. A small generator builds the book structure. Quarto outputs PDF, EPUB, & HTML.
dakoller.net/blog/2026082...
#Paperclip #Quarto #Markdown #Documentation #PDF #EPUB
dakoller.net/blog/2026081...
dakoller.net/blog/2026081...
The model install was only half the story: CUDA driver issues, VRAM contention, existing LLM workloads, and eventually a gpu-mode architecture for switching dynac between inference and video generation. dakoller.net/blog/2026081...
The model install was only half the story: CUDA driver issues, VRAM contention, existing LLM workloads, and eventually a gpu-mode architecture for switching dynac between inference and video generation. dakoller.net/blog/2026081...
I'm excited to share my latest project that solves a common problem at community events - collecting payments without making attendees download apps!
Blog post: dakoller.net/blog/2025101...
#SumUp #Payments #EventTech
I'm excited to share my latest project that solves a common problem at community events - collecting payments without making attendees download apps!
Blog post: dakoller.net/blog/2025101...
#SumUp #Payments #EventTech