[I wrote some of this up last week and I just keep forgetting to finish it and post it!}

So after all was said and done, we got probably two to three inches of snow a week ago. It’s hard to tell, because it snowed all day, changed to rain for a little while, then got cold enough to snow again.

I’m glad that I didn’t brush off the car last Monday (the 1st), it’s always easier in a snow/rain/snow storm to wait until everything is done to clear the cars. Because if you clear off the first layer of snow, all you end up with is a layer of ice with maybe some snow on top. And then instead of brushing off the car, I’m chiseling out the car.

***

Because I currently own two cars, I had twice as much snow clearing to do when the plow guy showed up. I NEED to dig out the title for the Honda, and clear the reenacting stuff out of the trunk so that I can donate it. I’d meant to do it before the snow started flying… But clearly I did not. And I don’t want to have to keep clearing off two cars. So time to buckle down and deal with the Honda.

***

One of the Engineering Managers had to duck out of a meeting last Wednesday (Dec. 3d) for a family emergency. It turned out that his 16 year old son, who just got his driver’s license earlier this year, had totaled the car. The kid’s fine, the car, not so much. He posted photos, and the kid skidded off the road and wrapped the front end of the car around a tree. EM said that if he’d had anything to say about it, the kid would not have been driving in the snow. But his mother (EM’s ex-wife) let the kid drive, and he was speeding. Personally, if I had a teenage boy who was still a novice driver, I would not have let him on the road during that particular storm, because it was a mix of rain/sleet, and sure to be slick out. I’m just glad the kid wasn’t hurt. It’s an expensive lesson though.

***

I finally got the second of my two “Docathon” projects into a pull request this Tuesday, a week after I got the first one into a pull request. I’m honestly really thrilled with what AI was able to pull out of the code base when I asked it to review all the error messages and write some troubleshooting documentation for how to resolve each error. It gave me much more detailed results than I was expecting. Since we’ve had lots of requests for improved troubleshooting, I’m excited to get this published for our users. I just have to nag my developers to review two big hoking documentation updates before I can get them merged and published.

***

After a lengthy conversation about cheese with my massage therapist a couple of weeks ago, Facebook showed me an ad for a cheese advent calendar from the Cheese Brothers. On a whim I bought two, one for myself and one for Jenny, my massage therapist. I couldn't wait, I cheated and peeked inside the box. The contents are:

* Hot pepper cheese curds
* Old Smokey smoked Gouda
* Honey sriracha Gouda
* Green onion Cheddar
* Chocolate chip cheese ball (Havarti)
* Reserve 8 year aged Cheddar
* Scorpion pepper Gouda
* Apple wood smoked Alpine style Adelheid
* Classic cheese curds
* Italian style Fratello (Inspired by Asiago and Parmesan)
* Parmesan Gruyere Yodel
* Dill Havarti

They all sound delicious. The only one I might not eat would be the Scorpion pepper Gouda, but that really depends on how hot it actually turns out to be. I plan on getting started on a couple of them this weekend. But as punishment for peeking, the cardboard box gave me a paper cut. It’s always the little boo-boos that hurt the worst. =(

I am so pissed off at one of our Documentation Managers right now.

She is a bit of a contrarian. She’s admitted that when someone says something, her first instinct is to jump in and argue for the opposite side. And sometimes that makes her especially annoying.

One of my European coworkers, who is also someone that I have trouble understanding, posted a blog post today. Now, some back story, when I say that I have trouble understanding her, it’s because even after three years, I barely know her. She misses most of our team meetings, because they fall right when she has to pick up her kids from school. OK, no problem. But we were actually on the same team for a bit, and she also missed a lot of those meetings. I finally wondered why I felt like I knew so little about her when we both work on database products and should be in a lot of the same Slack channels, yet I never see her posting. I went in to Slack and searched for all posts from her and it turns out that she only posts about once a week. And this goes back for months and months. So she’s a complete enigma to me, even though we’ve worked together for three years.

Back to coworker's blog post. She’s extolling the wonders of using an AI agent and how you can train it to think like you.

A documentation agent doesn’t wait for you to tell it what to do next. It’s not just “AI that writes with you.” It’s a system - a programmable version of your own brain - that already knows your process, your standards, your templates, your preferred IA patterns, and so on.

An agent in this sense is your AI twin, a codified version of your judgment, your craft, and your workflow, embedded directly in your repo, ready to activate any time a new feature, product area, or urgent request lands on your plate.

We’ve moved beyond the era of prompt engineering to the era of building AI systems - and for technical writers, that means creating your own documentation agent; a personalized AI twin that reflects how you think, how you structure information, and how you write.

She talks in very vague terms about training an AI, but gives no specific examples. So I asked in the Slack thread where she posted for some tips, or examples, or if I could see her copy of her “personal agent” file.

And I was told to just try it for myself.

Um, the reason why I’m asking is because it’s not clear how to do this myself! How is a Large Language Model supposed to know anything about how I learn, my 20+ years experience as a technical writer, or my personal workflows?

There was some back and forth conversation (thankfully a few other people seemed to be just as baffled as I was) and I finally posted

Other coworker: ”I honestly feel like [coworker’s] blog is further down the journey than I currently am, which is ok.”

Me: Same. I feel like sometimes when we're sharing info someone assumes that everyone else understands the skills they're using, when some of us haven't even been exposed to them yet. For example, how to get info into an AI chat. I can save things off the web that I want to dump into AI, it's easy to download Google Docs, but how are people getting things out of Slack and into VSCode or Cursor? Is it as simple as copy/paste into a file, ugly as that may be? Or is there some elegant trick that I just haven't learned yet? Same with issues, I haven't had a chance to play with issues yet. Do you have to download/save the issue, or can you just say "read issue 12345 and suggest how to resolve it"?

Manager: we've figured this stuff out by trying and experimenting though, not being taught by anyone... like just try stuff and see what works, then what works better

Me: How does refusing to tell me what you've learned help me here? How is it efficient for all of us we have to learn the same thing 20 separate times? This is the first week in a very long time that I've had time to play with AI, and I've gotten a lot done. But I haven't had time to try everything that I've wanted to. Saying "Just try it" doesn't find me time to do it.

Manager: because the most important learning is the journey not the result with AI

Which is utter bullshit, because we’re being pressured to produce results.

I've spent the past week experimenting with AI because it's Hackathon week, and that's what the docs team was working on. But I haven't had time to try everything because there's only so much time in a day.

The forecast a couple of days ago was three inches of snow.

The forecast Sunday night was 6 inches of snow.

The forecast Monday morning was 9 inches of snow.

The forecast Monday around dinner time was 2 inches of rain/snow mix.

The forecast as I’m going to bed Monday night is 1.5 inches of snow to rain. No, now it’s 2.2 inches. Wait, it changed again, now it's 3.5 inches. I swear, the forecast has changed every time I’ve thought to check it today.

I don’t know if this is a particularly tricky storm, or if the cuts to the National Weather Service have been that severe, but this feels a little bit ridiculous.