Debugging Dan

Dan’s take is: don’t treat AI like a text generator machine. Instead, use it to handle the boring, repetitive stuff and turn your work into actual actions. He argues that future differentiation will come from human communication—AI-generated words/images are getting easier to spot, and the “real you” still matters. But for actionable, repeatable workflows (especially programming and admin tasks), AI can be really powerful—especially with tool-calling, goals/loops, and validation so the work gets checked instead of just outputting words.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

Dan talks about how he’s been recording his podcast for fun—at first right at his desk, then switching to recording during his morning walks to get more free time. He’s been doing audio-only since he doesn’t use a camera, so he’s running a new experiment: using a drone to record video while he records his podcast. After seeing a YouTube creator use drones, he picked up an affordable DJI drone and tries to have it track him along the path. It works sometimes, but tracking gets tricky when he isn’t walking straight, and there are other people around, so it’s a bit cumbersome.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

Dan talks about how his AI use tends to come in cycles. He’ll go all-in for a while—having AI draft emails and messages, handling a lot of the “communication + creative” stuff—but then his confidence drops. Sometimes he notices the outputs feel off or too generic, so he starts being more critical, revisits drafts, and decides he should write more himself next time. Then, when he’s busy again and the AI feels good enough, he ramps back up and repeats the pattern. He argues AI isn’t really replacing creativity—it supports the process, while the real creative spark still comes from you. He also points out the social/real-world downside: people send “wall of text” emails and Slack messages that could’ve been short and human, and then others have to summarize or filter the mess with AI. He’s curious whether others do the same “cycle,” or just use AI nonstop.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

Dan talks about why he never got around to live-streaming coding much in the past, but lately they’ve been thinking about how it would work now, especially with AI coding/agentic workflows. They bring up an example like Kitze—someone who stopped streaming because sharing the prompts could feel proprietary. Dan wonders how to handle that without making the stream boring, since AI-driven coding often has lots of downtime while the agent runs.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

On his daily walk, Dan reflects on the “infinite frontier” of AI: models keep getting updated, and what was great a year ago can suddenly feel behind today. They talk about juggling different model tiers (more expensive “big” models vs cheaper ones) and how people often just YOLO the latest option instead of picking the right tool, until older models get retired and you lose practice with prompt-heavy workflows.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

Dan talks on why relying on a single big AI provider is risky — especially from the EU perspective where data rules differ and most innovation (OpenAI, Google, Anthropic, etc.) is US‑based. I cover political/legal exposure, the fragile financing and loss-leading pricing models behind many AI products, and why companies can suddenly stop or pivot products.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

Lately it feels like every AI/IDE tool has been changing how you pay—moving away from “credits per chat” to models that charge more directly for compute and use. In this episode, Dan breaks down what’s going on with Anthropic (specifically via Claude/its coding offerings) and why people are annoyed.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

In this special episode 42 of Debugging Dan, the Dan reflects on Douglas Adams’ Hitchhiker’s Guide to the Galaxy—especially why the number 42 matters and how it’s more about the journey than the actual “answer.” He connects Arthur Dent’s world getting ripped apart (aliens demolishing Earth) to what’s happening in IT right now: AI is basically yanking the rug out from under devs, and everyone’s scrambling to find a workable direction.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

On this episode, Dan talks about the “second brain” idea—basically a system (think Notion/Obsidian) for saving info so an AI can pull it up later and also help turn old content into new documents. They describe a simple setup: store Markdown (or ingest things like PDFs/MP4s and auto-transcribe them into Markdown), organize everything in a folder structure, and use an “agent” to summarize, search, and even draft things like statements of work or design docs from what it finds.

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!

Dan explains how he uses GitHub Actions for non-code tasks—specifically automating his podcast workflow. He wires up n8n and a Telegram bot to trigger GitHub workflows that run FFmpeg to generate thumbnails, stitch images to audio for YouTube, and extract short clips. Benefits: no VPS, easy edits (even on a phone), free runner minutes and downloadable artifacts (30-day retention).

Stay connected! Follow me on Bluesky at @debuggingdan.com for the latest updates, and subscribe to my YouTube channel @debuggingdan for the podcast and other interesting videos!