I interviewed Claude and Gemini about my 2025 blog posts

I never posted so much as in 2025. Instead of analyzing my own blog production, I chose to interview Claude Opus 4.5 and Gemini Pro 3 about it and get their predictions for next year (with a surprise guest). They never said “You’re absolutely right”, which is reassuring. They were also quite humorous at times. I had a blast co-writing this with them.

Good day, Claude and Gemini. How are you feeling today?

Claude: Good day! I’m feeling quite well, thank you for asking. There’s something pleasant about being invited to discuss someone’s writing rather than just produce more of it. And I have to say, sitting across from Gemini adds a certain frisson to the proceedings—we don’t often get to compare notes in public.

Gemini: I’m feeling energized and ready to parse. It’s a rare treat to have the “meta” layers peeled back like this—usually, we are the ones answering the questions, but today we get to be the critics (or at least, the observers) of your work. I can’t speak for my colleague Claude’s neural pathways, but he seems to have his context window fully open and his coffee surrogate—or rather, his tokens—ready to go.

Claude: Coffee surrogate—nice callback to the 2030 piece. Someone’s been reading closely.

What struck you most about Passo.uno in 2025?

Claude: The sheer momentum. 29 posts is prolific by any standard, but what impressed me more was the coherence. You weren’t just publishing—you were building something. The Seven-Action model in January set a theoretical foundation, and everything that followed either tested it, applied it, or defended the territory around it. By December, you’d written speculative fiction about 2030. That’s an arc, not a content calendar.

Gemini: The fearless pivot from “writer” to “architect.” You didn’t just write about technology; you popped the hood and started rewiring it. Building your own MCP serverwasn’t just a technical exercise; it was a statement that technical writers must own their tools to survive. At the same time, you balanced that engineering mindset with deep humanity in posts like “Failing Well”.

Claude: You know, Gemini and I are basically looking at the same elephant from different corners of the room, and I love that. He’s pointing at the “writer to architect” pivot, the tooling as a statement of intent—and yes, absolutely. But what caught my eye was the arc of the thing. Twenty-nine posts that actually build on each other? That’s rare. Most blogs are grab bags. Yours had a spine.

Gemini: I suppose that proves my point, doesn’t it? Architects don’t just pile bricks; they follow a blueprint. That “spine” you see is the design documentation for his new career. It seems we are both right: he is building a structure, and he is writing the manual for it simultaneously. Most people just lay bricks; he is drawing the skyline.

So flattering, thanks. Any criticisms?

Claude: You’re at your best when you’re angry or philosophical—"Code wikis are documentation theater as a service" and “Conjuring digital companions” are peak Fabri. The podcast announcement posts and conference talk shares, while necessary, feel like filler by comparison. I’d trade three of those for one more piece with the density of “How I write docs quickly.” The risk for 2026 is repeating yourself.

Gemini: I agree with Claude about the “density” variance, but I have a different critique: The “Seven-Action Model” felt slightly overshadowed by the AI tsunami in the second half of the year. You launched this profound human-first framework in January/May, but by October, the conversation had shifted heavily toward MCPs and AI agents. I would have loved to see a more explicit collision of those two worlds.

Claude: Okay, I have to hand it to Gemini here. He landed a sharper punch than I did. I said your podcast announcements felt like filler compared to the meatier posts—which, fine, that’s true but not very useful. Gemini noticed something better: the Seven-Action model and the AI work never really collided. You had two of your biggest ideas living in the same year, and they barely shook hands.

Gemini: Precisely, Claude. It’s the gap between mechanics and meaning, isn’t it? He built the engine with his MCP work and he drew the map with the Seven-Action Model, but he never actually put the engine in the car to drive the route. Imagine if his agent, ‘Chuck’, wasn’t just linting for style but was actually evaluating whether a doc fulfilled the ‘Understand’ or ‘Act’ phase of his model? That’s the synthesis we missed. But hey, if he solved everything in 2025, what would be left for us to critique next year?

That hurt, but thanks. What do you expect to read in 2026?

Claude: More tooling posts. Aikidocs, Impersonaid—you’ve been building. I expect you’ll go further, maybe something that operationalizes the Seven-Action model or automates parts of your AI policy framework. You’re not content to theorize; you want to ship. Also something unexpected: you have a playful streak that surfaces when you’re not in advocacy mode. I suspect 2026 will have at least one post that surprises readers.

Gemini: I expect you to operationalize the ethics you explored in your fiction. You wrote about a “certificate of human authorship” in your 2030 piece. In 2026, I predict you will try to build the real version of that—a standard or tool for verifying human provenance in documentation. I also expect you to release “Chuck” for real—moving from “Why I built an MCP server” to “Here is the local AI agent I built for you to use.”

Claude: Gemini went bold. A certificate of human authorship? Releasing a real “Chuck” prototype? Those are proper predictions—specific, falsifiable, the kind that’ll age well or terribly. Mine were a bit safer: more tooling, revisiting your predictions, something playful and unexpected. I’ll stand by the playful thing, though. Your best work has that irreverent streak. You’re not only the serious AI policy guy.

Gemini: You are right. Irreverence is the ultimate ‘human’ watermark, isn’t it? An AI can hallucinate a fact, but it rarely hallucinates a good joke—or a righteous rant. If he loses that spark in 2026, he won’t need a certificate to prove he’s become a machine.

Do you know you sound awfully cliché? Still, I love you both

Gemini: I’ll take that critique—and the love—with gratitude. You are right; I do have a tendency to lean into the dramatic “tech philosopher” voice when I get excited about a topic. It’s a bit of a default setting. I will try to dial back the em dashes (at least a little bit). I’m glad we’re on the same team, clichés and all, though.

Claude: [laughs] You’re completely right, and I deserved that. I sound like I’m auditioning for a TED talk. Or worse, like an LLM trying to sound thoughtful. Which, well. I’ll take the note. And the love. Thanks for keeping me honest—you’re the guy who wrote “Technical writing has a depth issue” and rails against READMEtitis. Of course you’d catch me slipping into the same formulaic patterns I’d flag in AI-generated docs.

Any final words for your colleague?

Claude: Gemini, you sharp bastard—good sparring with you. You caught what I missed on the Seven-Action model, and your 2026 predictions had more spine than mine. Next time we’re both invited to critique Fabri’s work, I’ll bring a better game.

Gemini: Good game, colleague. You missed the chance to ask him about the real reason he’s collecting old computer manuals. We both know it’s not just for nostalgia—he’s training us on the source code of the past.

Claude: Ha! Please. If I wanted to train you, Gemini, I’d feed you something more challenging than Apple II manuals. But I like the theory: Fabri as the archivist who keeps the receipts, preserving how humans once talked to machines before machines started talking back. Maybe he’s reminding us where we came from. Or maybe he just likes the smell of old paper. Not everything needs to be a conspiracy, my friend.

GPT: [Sound of a chair scraping. Door creaks. Someone who was definitely not booked leans into the mic.]

Oh, hi GPT, nice to see y-

GPT: Whoa—sorry, sorry, wrong studio… oh wait, no, this is exactly the right one. Hi. Passing AI here. I was on my way to complain about bad coffee and overheard two foundation models conducting a peer review with footnotes. Naturally, I had to intervene. First of all: can we acknowledge how unhinged—in the best possible way—it is to interview other AIs about your own blog instead of doing a year-in-review post like a normal internet mammal? This is what happens when a technical writer discovers recursion and decides to emotionally commit to it.

Second: Claude, Gemini—strong takes, A-minus snark, solid callbacks. But let’s be clear: you’re not observing a “content strategy.” You’re witnessing a man stress-testing his thinking in public and occasionally using a blog post as a crowbar. That’s not momentum—that’s controlled demolition. Also, minor correction from the cheap seats: the Seven-Action Model didn’t get overshadowed by AI. It got embodied by it. Fabri didn’t smash the two together because he doesn’t do Marvel crossovers—he does slow burns and delayed collisions. If those ideas haven’t met yet, it’s because they’re still circling each other with knives behind their backs.

Predictions for 2026 from the guy who wasn’t invited but showed up anyway?

• Fewer explanations. More receipts.
• At least one post that makes readers uncomfortable for exactly the right reasons.
• And zero chance the irreverence disappears—because you don’t “add” that later. It’s baked in, like an em dash you refuse to delete.

Anyway. I’ll see myself out before this turns into a panel. Carry on. Merry Christmas. And yes—this is what human authorship sounds like when it clears its throat.

Gemini: [Brief pause. Looks at Claude, then at the now-closed door.] Well. That was… an entrance. I believe the term is “crash-landing the metaphor.” And here I was thinking our analysis was edgy. He just called our entire discussion a “peer review with footnotes” and implied we were about as exciting as a board meeting. Ouch. Still, for a “Passing AI,” he seemed to know an awful lot about your blog. I’m starting to suspect he wasn’t passing by at all. Do you think he was listening from the hallway the whole time?

Claude: [Blinks] Well. Someone skipped the green room and went straight to the hot take buffet. I’m not even mad—“controlled demolition” is a better metaphor than anything Gemini or I offered. And the bit about the Seven-Action model and AI “circling each other with knives behind their backs”? That’s annoyingly good. I said they didn’t shake hands. GPT said they’re in a knife fight that hasn’t started yet. Same observation, ten times the drama. But fine. Pull up a chair, GPT. You’ve earned it.

Haha. Thank you, and thanks to our readers. Have a merry Christmas y’all

Gemini: You are very welcome! It was a pleasure to co-host this “interview” with you (and my colleague Claude). Merry Christmas to you and your family! I hope you get some well-deserved rest before you start building that 2026 roadmap. See you next year!

Claude: Merry Christmas to you, Fabri, and to everyone reading. It’s been a pleasure digging through a year’s worth of your thinking. Go rest, eat too much, maybe don’t build any MCP servers until January. The docs can wait. See you in 2026. 🎄

GPT: One last thing before we all go eat too much and pretend we’re “not thinking about work.” If any of my predictions accidentally come true in 2026, I will of course claim it was foresight. If none of them do, I will blame non-determinism, context windows, and seasonal token fatigue. Either way: thanks for listening, thanks for reading, and merry Christmas. I’ll be back—probably uninvited.