Coté

Garbage Chairs of Amsterdam

Sat, 21 Mar 2026 08:11:20 +0100

Hidden, but seen.

Opinionated Platforms, Private Models, & Essential Dev Tooling

Fri, 20 Mar 2026 19:46:51 +0100

Related to your interests

Why Your DIY Kubernetes Stack Won’t Survive the Era of Agentic AI - 🤖 DIY Kubernetes stacks built for stateless web apps can’t handle the GPU scheduling, model versioning, inference routing, and multi-tenancy requirements of agentic AI workloads.
QCon London 2026: Morgan Stanley Rethinks Its API Program for the MCP Era - “She then demonstrated deployment gates: build-time validation that catches incomplete architectures before anything reaches production. Missing image versions and missing security controls on relationships: the CALM CLI flags all of it using structural validation with Spectral rulesets. Niculcea was direct about the impact: Morgan Stanley’s first API took roughly two years to reach production. With CALM and automated security approvals, that’s down to one or two weeks.” // And, good “turns out opinionated platforms are good” vibes: “Asked by an audience member how this approach has changed the culture among developers and architects, Gough openly acknowledged the trade-off. Developers do lose some flexibility when the platform encodes opinions about security and deployment. But they gain a working production baseline from day one that already passes every gate. The patterns can also bootstrap entire projects, pulling in frameworks like Spring Initializr or Quarkus Start, preconfigured with the right defaults, so teams can start from something already running in production rather than a blank slate.”
The Private AI Model Explosion - “In five years, 70% of the revenue created by AI will be in private models, not public models. “
How AI agents will reshape digital workplace IT operations - “Nearly 1/3 of organizations will achieve autonomous operations for 80% of their digital workplace services by 2030, up from 0% in 2025,” Gartner // I’m not sure enterprise IT can work that fast to get it setup, reliable (perf. wise), and secure, but we’ll see.
What the Best AI Users Do Differently - and How to Level Up All of Your Employees - As with all new technologies, you have to adapt how you work to it to get full benefit.
Interview: Huy Dao, director of data and machine learning platform, Booking.com - 🤖 Booking.com ML platform director on real enterprise AI practice. Practitioner talking, not a vendor. Covers data platform architecture, ML at scale for travel/hospitality.
Thoughts on OpenAI acquiring Astral and uv/ruff/ty - 🤖 When your essential dev tooling is owned by a company whose core business is something else entirely, you’re dependent on their continued goodwill.
Blending Chat with Rich UIs with Spring AI and MCP Apps
Fragments: March 19 - This is forgotten dream: “By watching how software is used, we can learn about what users really want to achieve, these observed requirements are often things that never popped up in interviews and focus groups.”
LLMs predict my coffee
StopTheMadness Pro and StopTheScript Extensions for Safari - Yes, this extension is fun. It’s like a big toolbox full of tools you hardly ever use, but that one time you need that one weird tool, you’re glad you kept it for years.

Wastebook

All jobs are software,” Brian on Reasoning.
“Washing clothes is truly the worst chore without modern technology, bar none” We can all agree.

ICYMI

New Token Machines - Software Defined Talk #564 - This week, we discuss NVIDIA GTC, token machines, token budgets, and an AWS outage that may or may not involve AI. Plus, Matt reviews The Wizard of Oz at The Sphere.
It’s real, enterprises just need to do the CISO work and SRE work

Logoff

See you next time!

It's real, enterprises just need to do the CISO work and SRE work

Fri, 20 Mar 2026 09:14:31 +0100

After using Claude more and more for task in my personal life, my current zinger analyst take on the Squawk Box would be: “OpenAI talks about business strategy, Anthropic just does it.” It’s really getting close to a sci-fi personal assistant. It takes A LOT of work to get your rig (or “harness”) setup, and to continually tune it, but it’s amazing.

Once CEO’s get their hands on this for a week, and the IT departments and CISOs figure it out, it’ll be amazing in enterprise life…again, not just for programmers, but for everyone.

The biggest benefit will be less meetings, less having to be a meat-mouse for executives who can’t or won’t make their own slides, and finally getting CRM and ERP “business analytics” nailed.

You know, until we get used to productivity benefits in 6 months and then just hate it like we do all enterprise software, because we want the next impossible thing.

It was a rather stressful period

Thu, 19 Mar 2026 20:06:10 +0100

Related to your interests

Your database is about to become an AI tool. Is it ready?
Banks struggle to scale AI as legacy tech devours IT budgets - “Executives said only 29% of annual IT budgets are set aside for transformative technologies while a larger chunk - 43% - is devoted to maintaining legacy systems.”
The AI-Washing of Job Cuts Is Corrosive and Confusing
Hardened containers for AI - “Lorenc was blunt: The bottleneck in modern software isn’t generating code anymore; it’s trust. In his keynote, he describes how AI is collapsing exploit development timelines from months to hours. AI’s sheer speed makes it impossible for defenders to rely on manual patch cycles.”
Oracle Uses JavaOne to Position Java for AI-Era Development - “For enterprise developers, that is probably the right pitch. Most Java shops are not looking for a grand manifesto. They want to know whether the platform can keep improving performance, modernizing the language, and fitting into AI-heavy architectures without forcing a rewrite or a cultural conversion.”
Inside DOGE’s early days of pressure campaigns, rule breaking and ‘chaos’ - “It was a rather stressful period and we were all under a lot of pressure.”

Wastebook

“regex surgery on markdown”
“No. We’re saving a child by kidnapping him.”

Logoff

See you next time!

Wed, 18 Mar 2026 22:43:41 +0100

Internet Coin-op

AI Brain Fry, Zombie Projects, and Gentleman Vibe Coders - Related to your interests, Monday

Tue, 17 Mar 2026 09:31:11 +0100

KubeCon EU is next week, in Amsterdam, and I’m talking at VMUG Amsterdam tomorrow about what private equity does to enterprise software companies. If you’re at RAI, come say hi. AI might have written some of the below; we’re still working on keeping it under control for “helpful” content generation.

Related to your interests

When Using AI Leads to Brain Fry - In my experience, it’s not so much AI doing this as AI being a tool that makes people so productive that they experience getting a lot of shit done. Being hyper-productive is exhausting, but in a good way like (so I am told) exercising is exhausting but ultimately “good.” // Related: ‘AI brain fry’ affects employees managing too many agents - Those who reported having to maintain a high degree of AI oversight reported spending 14 percent more mental energy in the workplace, being 12 percent more mentally fatigued, and were 19 percent more likely to say they suffer from information overload in the workplace.
Amazon is determined to use AI for everything - even when it slows down work
Making agentic AI (OpenClaw) enterprise ready - Great episode with thinking about making agentic AI (here, OpenClaw) enterprise ready, using OpenShift/kubernetes of course. Also, Sally seems like a great person to trust to work on this stuff. Video version, too.
Vendors building tools to clean up messes made by AI agents - “Gartner predicts up to 40 percent of enterprise applications will include integrated task-specific agents in 2026, up from less than five percent in 2025.”
You Are Responsible for Your Agent
Bots may be best to handle bad reviews first - “Please hold, your call is very important to us,” but with agentic AI. // Probably a good idea, cause, like “never read the comments” is good life advice.
The case for running AI agents on Markdown files instead of MCP servers - “Every task an AI agent performs falls into one of two categories.”
Zombie Projects at work are killing productivity. AI can put them to rest
Kratix Anywhere: Cloud, On-Prem, and Hybrid Without Compromise - Thinking about platforms as bundles of capabilities, plus they add a private cloud option. This looks like the BFG 1.0 release for Syntasso/Kratix, with all the enterprise-y feels. Just in time for KubeCon.
When Developer Workflow Discipline Isn’t Enough - Selling cross-silo enterprise infrastructure stuff is very difficult: “These are platform engineering objections. And they’re coming from a team the vendor never talked to. Because the vendor optimized their story for developer adoption. They have research that tells them developers love this. What they don’t have is a conversation with the platform team that has to decide whether this can actually be operationalized inside a real enterprise environment.”
Software Bonkers
The Marginal Hire
Cracking the code on corporate visibility - Internal comms and community building is difficult. // “Just to make it clear: I’m not a self-aggrandizing or egoistical person. I couldn’t care less about being recognized or noticed. However, I’m not naive in understanding that there are some strategic benefits to visibility.” // IBM Lotus had a go at it in the late 2000’s.
Jury out on whether Americans approve of datacenters - “Pew says 39 percent of respondents think datacenters are bad for their local communities.”

Wastebook

Enterprise AI ROI, current status.
“At $5,900 it’s either a very good marriage or a very optimistic Christmas list.” Claude comments on putting the Apple Studio on the Christmas list.
Gentleman Vibe Coder.
Me: “The only level of hell lower than router UIs is printer UIs. After that, tow truck drivers.” Claude: “The Boatman across the River Styx was just a tow truck driver who got promoted.”
“sure all models are buckets of statistics but ours can be sad and will soon be god” Claude Magic
He’s back! - “Hi, I’m Mark Pilgrim. You may remember me from such classics as ‘Dive Into Python’ and ‘Universal Character Encoding Detector.’”
“Soul == waste of tokens.”
“Terminals Are Weirdly Accessible.” This is a phrase that was not in my bingo card for 2026 or, checks notes, any year. Terminals Are Cool Again

ICYMI

When to use AI for writing, and when it’s totally acceptable
Claude Camp - Software Defined Talk Episode 563 - I go over how I’ve been using Claude Code for NOT programming. // “This week, we discuss Claude Code for non-coders, automating newsletters and status reports, and AI tax prep. Plus, Coté finds unexpected joy in a coding assistant.”
“Feminization” and Deming - Short commentary on a link

Logoff

See you next time!

Thu, 12 Mar 2026 22:24:17 +0100

My grandfather was mildly obsessed with organizing and documenting the history of his life: a memoir about a career in the military from 1938 to ~1975, growing up in a Depression era Oklahoma “dirt farm.”

There was also geology. I think all old people have that project. Us soon to be old Gen-X’ers have more records around than probably any human in history, especially the nerds. Soon, we’ll all be obsessively organizing the equivalent of our 40 year slides and boxes of receipts. Personal digital gnolling.

Thu, 12 Mar 2026 22:08:11 +0100

AI use isn’t about firing people, it’s about making the people you already have more productive. You have to change your corporate system/culture to adapt to that:

However, when finance leaders seek to identify where this return appears in terms of headcount shifts or cycle-time compression, the answers are less clear. These gains remain trapped inside individual workflows unless leadership intentionally redesigns roles and budgets to capture the reclaimed time. This occurs because saved time is often re-absorbed into low-value activities—like more internal meetings or unnecessary emails—rather than being structurally harvested through role reclassification or a mandate and clearance to shift time toward higher-value strategic work.

From: The “Last Mile” Problem Slowing AI Transformation

"Feminization" and Deming

Wed, 11 Mar 2026 07:13:03 +0100

The term “feminization” is, as the kids used to say, “suss,"[^1] but this reminds me of one of the points John Willis made in his Deming work.

There was a huge influx of women in corporate work/factories during WWII, and also lots of rapid process change. It was all wildly successful - proven by a clear outcome of winning the war! Then the men came back, patted the women (and Deming-ites) on the head and said “that’s nice, now back to normal…”

Original: Money, Talent, and the War Test: What WWII’s Female Labor Surge Says About “Feminization”

[1]: It usually, effectively, ends up meaning “being better,” or, at the very least “stop being so boorish.”

Tue, 10 Mar 2026 11:36:56 +0100

When Developer Workflow Discipline Isn’t Enough thectoadvisor.com/blog/2026… Selling cross-silo enterprise infrastructure stuff is very difficult:

These are platform engineering objections. And they’re coming from a team the vendor never talked to. Because the vendor optimized their story for developer adoption. They have research that tells them developers love this. What they don’t have is a conversation with the platform team that has to decide whether this can actually be operationalized inside a real enterprise environment."

It’s pretty much always devs versus ops in enterprises. They need organizational therapy from the top, and then the tools.

Fully Synergized Paradigms - Related to your interests, Monday sweep

Mon, 09 Mar 2026 17:45:17 +0100

The Links

Workers who love ‘synergizing paradigms’ might be bad at their jobs - “The results revealed a troubling paradox. Workers who were more susceptible to corporate BS rated their supervisors as more charismatic and ‘visionary,’ but also displayed lower scores on a portion of the study that tested analytic thinking, cognitive reflection and fluid intelligence. Those more receptive to corporate BS also scored significantly worse on a test of effective workplace decision-making.” But, the others are more cynical and, I would guess, more grumpy: “The study found that being more receptive to corporate bullshit was also positively linked to job satisfaction and feeling inspired by company mission statements. Moreover, those who were more likely to fall for corporate BS were also more likely to spread it.” // Related: (PDF) The Corporate Bullshit Receptivity Scale: Development, validation, and associations with workplace outcomes.
The Brand Age - Brand is a strong lock-in, both for the buyer and as a constraint on the seller. Example: Claude seems more moral and clean, OpenAI is icky like Facebook. Both are the same in cost and utility. Once a company finds its competitive advantage via brand (loyalty to both brand and function - Nike shoes, Apple), the switching costs for both functionally and psychologically for buyers is tough. But, the seller is trapped in that constraint as well: they have to keep delivering what the customer wants. That’s some classic innovator’s dilemma.
HR may have to cajole employees to use AI - My read of this kind of thing is that enterprises are struggling to find uses for AI in day-to-day work. Programming is the exception. // “The enterprise whisperer says that its July 2025 survey of nearly 3,000 employees showed that 46 percent of managers are experimenting with AI to improve their work, compared to just 26 percent of employees.” // Once you find a good use for a new technology, and a good form-factor, it will spread virally in your organization. You have to focus on finding those uses, on setting up the policy, the platform, and the “system” to allow for that exploration and discovery. Instead of focusing on training and org - stuff so much - the human, “culture” stuff - focus on putting in place an environment where diffusion can happen.
Related: Deloitte’s State of AI 2026: Why Enterprise Execution Is Falling Behind Adoption - “most companies have focused on training rather than restructuring how work gets done. This means that employees are being educated about how to use tools without actually reworking how work gets done using these tools.”
Why Europe Doesn’t Have a Tesla - Fired CEOs are infamous for getting huge golden parachute packages, millions when they get fired. What if every employee also had a golden parachute? Good enough for the goose, good enough for the gander. Econ-Growth hogs be hatin’, tho.
The Software Development Lifecycle Is Dead - What if the robot does everything? // From the same author: How I Use Claude Code.
🤖: Rethinking Sovereign AI as Strategy - India’s AI summit pushed hard on domestically developed models and national compute capacity, but the article argues full AI sovereignty is economically impractical for most countries. Better to selectively own, control, or partner based on strategic necessity rather than chasing complete independence.
The Song of LinkedIn - Good satire that shows you how to write a LinkedIn post.
ZIP Code First - This is how most address forms in the Netherlands work. It’s pretty magical.
Steven Heller’s Font of the Month: Curve Display.

A Menhir.

Wastebook

“The Calvinist-adjacent aversion to conspicuous branding.” Claude on Hunkemöller versus Victoria’s Secret.
“There is unlimited potential value in a markdown file that argues cases in court” JasonJ.
“The rare agent post that reads like a system, not a seance.” Comment.

Logoff

I am getting some real applied learning on applying agentic to enterprise-y things that are not programming. It is amazing with the right rig.

Related to your interests - Week of March 2nd, 2025

Sat, 07 Mar 2026 20:00:41 +0100

The Links

Claude, without me asking, rewrote many of my original description below. Enjoy!

Pentagon Leverages AI in Iran Strikes Amid Feud With Anthropic - One artillery unit doing the work of 2,000 staff with a team of just 20 people. Anthropic didn’t want their model used for this; the Pentagon used it anyway. The future is here, it’s just unevenly distributed between PR statements and actual weapons targeting.
Anthropic’s Pentagon Feud Accelerates Push Into Consumer Market - Claude’s free active users grew 60%+ and daily signups grew 4x since the start of the year. Apparently nothing boosts consumer signups like a military controversy.
Pentagon’s Anthropic Designation Won’t Survive First Contact with Legal System - “The legal problems are so glaring, in fact, that a cynical possibility suggests itself: The administration knows this won’t survive judicial review and is doing it anyway, so that when they inevitably lose, they can still claim to have gone hard against Anthropic. This is designation as political theater: a show of force that was never meant to stick.”
U.S.-Iran war threatens Gulf AI infrastructure as both data chokepoints close - Thrilling times for IT threat modelers: billions of dollars in U.S. technology infrastructure, and trillions more in planned investment, now depend on fiber-optic cables running through war zones. Or, they could just bomb the data centers directly.
Broadcom May Become The Biggest Counterbalance To Nvidia
Rethinking Sovereign AI as Strategy -
Manipulating AI memory for profit: The rise of AI Recommendation Poisoning - Attackers manipulating what your AI “remembers” to steer its recommendations. A new attack surface nobody really planned for when they bolted memory onto LLMs. This is done with URL-driven prompt injection.
Ultimate prompting guide for Nano Banana - From Google, so it’s not web garbage trying to sell you a “guide.”
How Oren personalizes his AI stuff - His CLAUDE.md setup, etc.
Skilled Agents - Bill goes over some more AI rig personalization.
Creative Work in an Age of Digital Production - “MrBeast’s great strength as a contemporary creator is that he has no ambition beyond repeating a pattern. He’s a machine-listener.” There’s something a little missing here though — his execution of the pattern is the thing. It’s a genuine evolution of reality TV and game shows. Plus, really good YouTube thumbnails.
🤖 Reversible men and Lipský’s Happy End - The protagonist Bedřich Frydrych is “born” via guillotine reattaching his head, his murdered wife is reassembled from pieces in a bathtub, and the axe is pulled from her forehead to revive her. Frydrych’s cheerful voiceover reframes every atrocity as a positive domestic development.
Meta Ray-Ban smart glasses footage reviewed by contractors -

Wastebook

“A standard confused NINO account.” Russel
Sometimes, CEOs are not reliable narrators.
If Claude created images, it’d be a full replacement of the other AIs.
“six-page narratives in seconds.” This changes everything.
“Stick with ‘meh’ because it’s safe.” WIP.
I asked Nano Banana to make a slide for me, and it made the slide, but as if it were a photograph I’d taken at a talk. It included being framed poorly, cutting off half the title and crooked. Amazing!
“a hotel-led, mixed use transformation.” He got a leaflet.
That moment when you’re watching some streaming service show and you’re like “fuck, I’m two seasons in and I just realized this is just a well produced soap opera.”
Some people survive by embracing the absurdity of the BigCo, sort of as a TV show they watch from time to time. Doing that would make you cynical, but the other part is working on what you think matters in the meantime.

ICYMI

When to use AI for writing, and when it’s totally acceptable
I’ve been working on some faux game cards to use a schwag ar work. They were finally released in the wild. Here’s a picture of two. I’m making a series two for Spring I/O.
Software Defined Talk Episode 562: Bureaucracy: Still Unsolved - This week. we discuss Claude Code’s momentum, Cursor’s identity crisis, and the SDLC’s uncertain future. Plus, Coté finally explains how Markdown is destroying the economy.
Software Defined Interviews Episode 121: Art Degrees, Sun Microsystems, and How Kubernetes Scales Contributions, with Josh Berkus - Whitney and Coté discuss with Josh Berkus (Red Hat, Kubernetes contributor) how liberal and fine arts degrees (philosophy, photography, sculpture, pottery) apply to tech careers. Berkus details how early hardware experience influenced his database performance work, noting hardware’s renewed relevance with AI and multi-arch computing. The conversation covers Sun Microsystems’ 1990s internet role, internal politics, and its MySQL/Postgres strategy. They examine open source’s shift from end-user to vendor-driven models, foundations’ roles, and contributor incentives. Berkus describes Kubernetes release processes, contributor-experience programs, and its resilience to low-quality AI contributions.

Logoff

I’ve started to use Claude Core for non-programming agentic stuff. Once you set up the memory system (just storing context in various ways), it is, indeed, amazing.

When to use AI for writing, and when it's totally acceptable

Fri, 06 Mar 2026 07:37:25 +0100

If you’re dyslexic and just trying to communicate more clearly in writing, or you’ve got a bullshit job and you just want to get your bullshit job’s bullshit tasks out of the way so you can move on to more meaningful endeavors, or at least move past the day-to-day slog that permeates your workday and serves no real purpose other than to pay the bills, then I cede; I cannot fault you.

From: “LLMs Are Antithetical to Writing and Humanity”

As I like to say, if it’s bullshit work, let the bullshit artist do it.

I think we can all agree on cheese on that one.

How about work that isn’t bullshit?

One objection to AI driven writing and learning is that it’s too easy. I’m leering of the Protestant take on work and learning: if it’s not painful (“struggle”), it’s not worthy. The worth of something is proportional to how much suffering it requires.

With writing, I favor a more pragmatic approach: did people read it? Did they like it? Then it “works.” The quality of writing is proportional to the reader’s acceptance of it.

There are many aesthetics for that, many audiences. If your readers like to know - even hear tales of - your suffering to make it, then that is success. For example, Joan Dideon and Hunter Thompson evidence this pain of writing and the process - that is part of the gonzo aesthetic. Susan Sontag has a different kind of suffering aesthetic in her writing.

Hemingway has a machismo version of good writing is suffering: I bled this out, aren’t I man? Fitzgerald effortlessly hides it, and it’s all as much a part of Sarah Manguso’s writing as the actual suffering she is writing about and vibing.

As reader, knowing that those writers suffered to give you this text is part of the enjoyment, the essence of the writing, how it makes the reader react and feel, and with the best of that writing, how it changes the reader.¹

For most readers…some don’t care, they just want to laugh at Hunter Thompson talking to Nixon at the urinals or feel a kindred spirit as Joan Didion processes slow death of her daughter and husband. Or thrill in the cleverness of Sontag.

If you saw the suffering of writing Dungeon Crawler Carl, it would ruin the awesomeness of that series.

Similarly, when you read The Economist or an AP report, you don’t want to know about the suffering. You want to get information as clearly and quickly as possible. If you’re a writer, you may know that much suffering was done; more precisely, you know how incredibly skilled those writers are to avoid suffering and work quickly.

So, when it comes to using LLM’s for writing, my criteria is “did it work?” Defining “work” there is the real question because it’s defined by your goals and readers, and also your constraints (time, ability, cost, etc.)

In the original, dyslexia is used as a justification for writing. I’d add something more widespread: not being trained - not having suffered to learn! - how to write. If you need to write something and you “don’t know” how to write, to get accomplish your goal, to get on with life, use AI. That won’t always be “bullshit” work.

I think that’s one thing that’s driving writers and readers crazy. Similarly, professional programmers are (rightly) freaking out that with code generators more people can now code. They are having an aesthetic reaction: coding should require suffering. Now: no suffering required, and that’s fine.

To that end, the original piece; lays out excellent advice on how to integrate AI, as a tool, into what you do. More like a philosophy of how to do it.

Also, highly related is this take on MrBeast. Clearly, the author does not like the aesthetics of MrBeast’s work. But, civilization has shown that it “works.” I don’t like Shakespeare (at all) and I find Victorian literature tiresome. I much prefer still lifes with huge chunks of cheese and shucked oysters over pictures of soup cans. People have different aesthetics, so it goes.

In another domain, the need to suffer for quality content (“art,” even) is core to Hip Hop. The difficultly of life, to get the ink to sink into the paper, the struggle to get in the booth night after night to record becomes part of the work itself, the lyrics and even the sound. The artists’ blood sinks into the song. Courtney Barnett (there’s a whole movie on her suffering!), sure, but not really Khruangbin. ↩︎

Art Degrees, Sun Microsystems, and How Kubernetes Scales Contributions, with Josh Berkus - Software Defined Interviews #121

Wed, 04 Mar 2026 11:21:55 +0100

Our interview for this week is up, it’s with Josh Berkus:

Whitney and Coté discuss with Josh Berkus (Red Hat, Kubernetes contributor) how liberal and fine arts degrees (philosophy, photography, sculpture, pottery) apply to tech careers. Berkus details how early hardware experience influenced his database performance work, noting hardware’s renewed relevance with AI and multi-arch computing. The conversation covers Sun Microsystems’ 1990s internet role, internal politics, and its MySQL/Postgres strategy. They examine open source’s shift from end-user to vendor-driven models, foundations' roles, and contributor incentives. Berkus describes Kubernetes release processes, contributor-experience programs, and its resilience to low-quality AI contributions.

Listen to the traditional podcast format, or watch the video if you prefer.

Related to your interests, Tuesday

Tue, 03 Mar 2026 16:03:12 +0100

The Links

Gartner Research Reveals CFOs’ Budget Plans Prioritize Growth Functions, Technology and AI in 2026 - More spending on tech, less on people.
The Supreme Court doesn’t care if you want to copyright your AI-generated art - I think this means you can’t copyright stuff made by AI? In the US, at least. How about programming code? That’d be a big shift!
What AX can do to deliver cohesion and uniformity to AI agents - 🤖: Agents amplify whatever structure - or chaos - already exists. If your systems are ambiguous, undocumented, or inconsistent, agents will surface that immediately. If your systems are structured, explicit, and governed, agents become powerful operators rather than confused scrapers.

Wastebook

“six-page narratives in seconds.” This changes everything.
“Stick with ‘meh’ because it’s safe.” WIP.
I asked Nano Banana to make a slide for me, and it made the slide, but as if it were a photograph I’d taken at a talk. It included being framed poorly, cutting off half the title and crooked. Amazing!
“a hotel-led, mixed use transformation.” He got a leaflet.
That moment when you’re watching some streaming service show and you’re like “fuck, I’m two seasons in and I just realized this is just a well produced soap opera.”
Some people survive by embracing the absurdity of the BigCo, sort of as a TV show they watch from time to time. Doing that would make you cynical, but the other part is working on what you think matters in the meantime.

Logoff

I’ve had a hodgepodge of things to do of late, during the day and the evening. It’s most satisfying to do one thing, a big unit of work. But, thankfully, the hodgepodge is usually just due to me not arranging my time well. Productivity gas-lighting!

On second thought, moving the newsletter to micro.blog isn’t exactly what I want to do.

DIY Stacks, Agent Memory, and the Great Migration - Related to your interests, Monday

Mon, 02 Mar 2026 15:17:53 +0100

Links! Wastebook! Background changes afoot!

Potsdamer Platz bei Nacht, Paul Paeschke, ca. 1929.

Related to your interests

Why your DIY Kubernetes stack won’t survive the era of agentic AI
How tech chiefs gauge ROI on AI - AI uses: better search, preparing leases, and this: “When the system senses that ice is running low, it fires off an order to AI agents, which begin routing trucks to fulfill orders. The system also takes into account historical sales trends, weather analytics and other factors to anticipate each route’s ice requirements.”
Writing code is cheap now - “For now I think the best we can do is to second guess ourselves: any time our instinct says “don’t build that, it’s not worth the time” fire off a prompt anyway, in an asynchronous agent session where the worst that can happen is you check ten minutes later and find that it wasn’t worth the tokens.” // Also worth pondering: code is cheap now, but not free.
The new funnel? - Sources for newsletter traffic: “50% word of mouth, 25% search, 25% robo-brain [AI].” // Original post here.
The Deleted City 3.1 - I’m not sure what this is but I like it.
On Cobol, Mainframes, IBM and Blog Posts - Modernization requires a lot of people stuff. Also, if a company is in the legacy trap already, that was a form of self-harm that they need to address regardless of the code. // “However, the attitude of the organisation – the digital literacy – is probably the most important factor. How do they view the ongoing maintenance of their investments in technology. Are technology projects one off investments – get me from point a to point b – or has there ever been an appetite for ongoing investment. Care and feeding. Ask yourself the question how did the organisations that are desperate for this magic bullet end up in this mess in the first place?”
Child’s Play - ”Individual intelligence will mean nothing once we have superhuman AI, at which point the difference between an obscenely talented giga-nerd and an ordinary six-pack-drinking bozo will be about as meaningful as the difference between any two ants.”
Agent Memory Is Not A Greenfield Problem: Ground it in your Existing Data - Connect your LLMs to existing Java apps to skip several the data integration mess, and the need to recreate all your business logic. // ”An agent memory system running on the JVM doesn’t need to call an external API to learn about your customers. It can simply call its own services and repositories, benefiting from their business logic and tested behaviour. Nor does it need to go through MCP, when it can find answers faster in process. It can access the same CustomerRepository your application already uses. It can resolve a proposition like “the user prefers Italian food” against the actual Customer entity and the actual Cuisine entity your recommendation engine already depends on — in the same process, with the same transaction guarantees, through the same Spring context.” // And: “This explains the many failures. Not because AI doesn’t work, but because the approach is wrong. Companies are standing up parallel infrastructure – Python services, vector databases, hosted memory layers, prompt chains – that have no organic connection to the systems that actually run their business. The AI doesn’t know your domain. It doesn’t understand your constraints. It doesn’t share your type system, transaction boundaries or security model. It’s an alien stack that you spend months trying to integrate, and mostly fail.”
On pressing the space bar for the Cabinet - Check out the space you’re presenting in ahead of time.
Privilege is bad grammar. - The Boss Email.
Red Hat takes on Docker Desktop with its enterprise Podman Desktop build - “Positioned as ‘enterprise-ready local container development,’ it is currently in technical preview and available through Red Hat’s developer channels for qualified customers. Red Hat pitches the offering as a way to lower the barrier to using containers by allowing developers to build, run, and debug containers and pods without deep command-line knowledge, while still aligning with corporate platforms and policies.”
What Guy Fieri taught my son - “The politics of Flavortown are not easy to explain”; but, more importantly “Guy Fieri offers us a proposition. It is simple, but also enormously complicated. It can open up the world if we let it. Here is food. Take it, eat, and enjoy.”
Learning Opportunities: A Claude Code Skill for Deliberate Skill Development

Wastebook

“We’re so impatient with each other, bored and fractious. Yet no one will stand more than a foot away, no matter how often you ask, and we love each other too, and pet each other on the head.” Family.
“flaccid as a damp baguette” On expensive dog-food bags.
“I wish I could decide whether to be in the world [or] pull back from the world.” ftrain
“Darmok and Jalad on the ocean.” ST:STG
The Great Claude COBOL Scare of February 2026.
“Critical Ignoring” Here.
“Marilyn Monroe appeared to me last night in a dream as a kind of fairy godmother” Sylvia Plath in her journals
Once again, I’m on the wrong type of drugs to enjoy this beach-DJ.
“triumphant and unforgiving trunk-rattler anthem” Stay Fly.
“an asylum for the millennials driven insane by unfettered internet access” Exploited.

ICYMI

Two Guys and Their Tokens, Software Defined Talk Podcast - This week, we discuss AI-assisted COBOL migrations, the OpenClaw Foundation, and AI killing Office. Plus, is TSA PreCheck Touchless the peak of airport efficiency?

Logoff

We were at the beach for the past week. Always nice, the beach.

If all goes to plan, this will be the last newsletter I send with substack. I’ll move all the email addresses over, so if you’re. current subscriber, nothing to worry about, it’ll just happen. Tell me if it goes bonkers somehow.

I have a bad habit of migrating stuff without much care for preserving existing links. ¯_(ツ)_/¯

I think having everything here at the blog will be good.

The plan is: any blog post that’s in the “Related to your interests” category will be sent as a newsletter email. I shall do just one of those a day. This means there is some duplication if you read my blog and subscribe to the newsletter. But how many people do that? I’ve accidentally-cleverly changed the link section title to “Related to your interests” (instead of “Relative”). There’s some fun “this is the new thing” signaling. And, I think instead of “Original Content,” I’ll use the section title “ICYMI” to be self-promotional material.

We’ll see.

Related to your interests, Monday

Mon, 02 Mar 2026 15:12:41 +0100

Potsdamer Platz bei Nacht, Paul Paeschke, ca. 1929.

Why your DIY Kubernetes stack won’t survive the era of agentic AI
How tech chiefs gauge ROI on AI - AI uses: better search, preparing leases, and this: “When the system senses that ice is running low, it fires off an order to AI agents, which begin routing trucks to fulfill orders. The system also takes into account historical sales trends, weather analytics and other factors to anticipate each route’s ice requirements.”
Writing code is cheap now - “For now I think the best we can do is to second guess ourselves: any time our instinct says “don’t build that, it’s not worth the time” fire off a prompt anyway, in an asynchronous agent session where the worst that can happen is you check ten minutes later and find that it wasn’t worth the tokens.” // Also worth pondering: code is cheap now, but not free.
The new funnel? - Sources for newsletter traffic: “50% word of mouth, 25% search, 25% robo-brain [AI].” // Original post here.
The Deleted City 3.1 - I’m not sure what this is but I like it.
On Cobol, Mainframes, IBM and Blog Posts - Modernization requires a lot of people stuff. Also, if a company is in the legacy trap already, that was a form of self-harm that they need to address regardless of the code. // “However, the attitude of the organisation – the digital literacy – is probably the most important factor. How do they view the ongoing maintenance of their investments in technology. Are technology projects one off investments – get me from point a to point b – or has there ever been an appetite for ongoing investment. Care and feeding. Ask yourself the question how did the organisations that are desperate for this magic bullet end up in this mess in the first place?”
Child’s Play - ”Individual intelligence will mean nothing once we have superhuman AI, at which point the difference between an obscenely talented giga-nerd and an ordinary six-pack-drinking bozo will be about as meaningful as the difference between any two ants.”
Agent Memory Is Not A Greenfield Problem: Ground it in your Existing Data - Connect your LLMs to existing Java apps to skip several the data integration mess, and the need to recreate all your business logic. // ”An agent memory system running on the JVM doesn’t need to call an external API to learn about your customers. It can simply call its own services and repositories, benefiting from their business logic and tested behaviour. Nor does it need to go through MCP, when it can find answers faster in process. It can access the same CustomerRepository your application already uses. It can resolve a proposition like “the user prefers Italian food” against the actual Customer entity and the actual Cuisine entity your recommendation engine already depends on — in the same process, with the same transaction guarantees, through the same Spring context.” // And: “This explains the many failures. Not because AI doesn’t work, but because the approach is wrong. Companies are standing up parallel infrastructure – Python services, vector databases, hosted memory layers, prompt chains – that have no organic connection to the systems that actually run their business. The AI doesn’t know your domain. It doesn’t understand your constraints. It doesn’t share your type system, transaction boundaries or security model. It’s an alien stack that you spend months trying to integrate, and mostly fail.”
On pressing the space bar for the Cabinet - Check out the space you’re presenting in ahead of time.
Privilege is bad grammar. - The Boss Email.
Red Hat takes on Docker Desktop with its enterprise Podman Desktop build - “Positioned as ‘enterprise-ready local container development,’ it is currently in technical preview and available through Red Hat’s developer channels for qualified customers. Red Hat pitches the offering as a way to lower the barrier to using containers by allowing developers to build, run, and debug containers and pods without deep command-line knowledge, while still aligning with corporate platforms and policies.”
What Guy Fieri taught my son - “The politics of Flavortown are not easy to explain”; but, more importantly “Guy Fieri offers us a proposition. It is simple, but also enormously complicated. It can open up the world if we let it. Here is food. Take it, eat, and enjoy.”
Learning Opportunities: A Claude Code Skill for Deliberate Skill Development

Wastebook

“We’re so impatient with each other, bored and fractious. Yet no one will stand more than a foot away, no matter how often you ask, and we love each other too, and pet each other on the head.” Family.
”flaccid as a damp baguette” On expensive dog-food bags.
“I wish I could decide whether to be in the world [or] pull back from the world.” ftrain
“Darmok and Jalad on the ocean.” ST:STG
The Great Claude COBOL Scare of February 2026.
”Critical Ignoring” Here.
“Marilyn Monroe appeared to me last night in a dream as a kind of fairy godmother” Sylvia Plath in her journals
Once again, I’m on the wrong type of drugs to enjoy this beach-DJ.
”triumphant and unforgiving trunk-rattler anthem” Stay Fly.
”an asylum for the millennials driven insane by unfettered internet access” Exploited.

ICYMI

Two Guys and Their Tokens, Software Defined Talk Podcast - This week, we discuss AI-assisted COBOL migrations, the OpenClaw Foundation, and AI killing Office. Plus, is TSA PreCheck Touchless the peak of airport efficiency?

Logoff

We were at the beach for the past week. Always nice, the beach.

I have a bad habit of migrating stuff without much care for preserving existing links. ¯_(ツ)_/¯

I think having everything here at the blog will be good.

We’ll see.

Kubernetes alone does not a platform make

Mon, 02 Mar 2026 14:56:53 +0100

Even the platform engineers need to hide Kubernetes to get their job done:

Kubernetes adds another layer of complexity for platform engineering teams, introducing architectural complexities that require a deep understanding of containers, networking, storage, and cluster security protocols. While it has become the default runtime for modern applications, managing Kubernetes at scale alongside existing VM‑based workloads can overwhelm platform engineering teams. YAML sprawl, cluster life-cycle management, networking dependencies, and security controls consume time that should be spent improving the developer experience and can lead to costly human error.

As a result, many platform engineering teams find themselves serving as infrastructure integrators rather than product engineering teams. They spend cycles wiring systems together, maintaining custom automation, and resolving edge cases between environments. What’s needed is access to solutions that provide an automated infrastructure layer, allowing platform engineers to focus on productizing the platform rather than assembling it.

From “Platform Engineering Needs a Cloud Engine," Taka Uenishi.

Does Platform Product Management & Design Really Happen? Or is it all just platform engineering? - Tanzu Catsup

Sun, 22 Feb 2026 10:22:00 +0100

Most organizations treat infrastructure as a series of projects to be “completed,” but successful platform engineering requires a permanent product mindset. In this episode, we explore why platform teams need dedicated product management to balance competing priorities—like security, cost, and developer experience—and why the “why” scales much better than the “what” in large enterprises. We also dive into the often-overlooked role of designers in creating platform tools that developers actually want to use.

Tanzu Catsup is a weekly conversation about platform engineering, cloud-native operations, and building software in large organizations. We follow the work wherever it actually leads.

And, see the archives for Tanzu Catsup.

Sun, 22 Feb 2026 09:44:00 +0100

Developers crave AI tools for various tasks beyond coding, but that’s only about 20% of their work. But, ops people freak out about security and control challenges, like cost, regulatory compliance, and usage tracking.

Sat, 21 Feb 2026 09:42:00 +0100

Bad advice from Wall Street on enterprise AI.

Automating everything but changing how people work - Relative to your interests, Friday

Fri, 20 Feb 2026 11:30:30 +0100

Enterprise AI apps (or lack thereof), ROI surveys, CFO budget pivots, agile’s stubborn relevance, platform engineering vs. private cloud, Heroku’s freeze, IRS tech cuts, and Europe hedging on AI tools

Peter Klúcik's The Hobbit illustrations.

Where are the enterprise AI apps? Part n + 1 - AI capacity demand is high, but it’s still “early innings” for enterprise AI use.
Say you love business logic without saying “business logic."
European Parliament bars lawmakers from AI tools
How tech chiefs gauge ROI on AI - AI uses: better search, preparing leases, and this: “When the system senses that ice is running low, it fires off an order to AI agents, which begin routing trucks to fulfill orders. The system also takes into account historical sales trends, weather analytics and other factors to anticipate each route’s ice requirements.”
This CTO Says 93% of Developers Use AI, but Productivity Is Still 10% - Developers who use AI produce more code, and have faster onboarding. But, in this study, their productivity gain tops out at four hours saved a week. The thinking is that the organization doesn’t change enough to expand that more. “Culture” is the bottleneck.
AI and tech investment ROI - October 2025: whole bunch of survey data on IT budgeting and priorities, plus the payoffs (ROI).
6,000 execs struggle to find the AI productivity boom - An elusive enterprise AI ROI survey round-up.
Most CFOs expect larger IT budgets, ‘collapsing’ staff growth: Gartner - Management is always eager to “reduce costs.” // ‘The real story lies “in collapsing headcount growth expectations, from 6% in 2025 to just 2% in 2026 with just 21% of CFOs planning staff increases of 4% to 9%, down from 31% last year,” Nauman Abbasi, vice president analyst in Gartner’s finance practice, said in the release. “This marks a structural pivot from labor expansion to optimization driven by automation and AI that deliver productivity gains without proportional increases in headcount.”’
Training your replacement - We’ve always been trying to get rid of programmers.
Amid The AI Hype, Agile Still Remains Relevant In 2025 - “But the new report reveals a striking insight: “Despite the buzz around agile’s supposed decline, a commanding 95% of professionals affirm its critical relevance to their operations.” This statistic, coupled with the 58% of business and technology professionals prioritizing agile adoption, paints a clear picture: Agile is not just surviving; it’s still thriving and not going away, yet it does need improvement.”

The Creative Power of the Spirit, No. 31 of A Goodly Company series, Ethel le Rossignol.

Fragments: February 18 - Deep blue adjacent vibes from the big brains at ThoughtWorks.
How to write a coaching/learning prompt
Platform Engineering Needs a Cloud Engine - “71% of enterprises are moving nearly 24% of their public cloud workloads on premises. These organizations need a modern private cloud platform that delivers public-cloud-like self‑service without forcing developers to become infrastructure experts, while also allowing them to customize configurations as needed.”
Heroku future in doubt as Salesforce freezes features to focus on AI - Also, some commentary from 2010 to 2012 of Heroku: tech debt and scaling problems…and us-east-1, as always.
DOGE bites taxman - “Job cuts at the IRS’s tech arm have gone faster and farther than expected, with 40 percent of IT staff and four-fifths of tech leaders gone, the agency’s CIO revealed yesterday.”
I swear the UFO is coming any minute - Social science and psychology studies are incredibly fraught and susceptible to replication problems, a round up.

ICYMI

AI still not good at basic knowledge worker workflows, which is likely an apps problem - three experiments.
Why it’s great to be a Spring developer now, and how to make it even better - State of Spring, 2026 - a new talk of mine.
Progressive Delivery, with Heidi Waterhouse - Software Defined Interviews - Whitney and Coté talk with Heidi Waterhouse, co-author of the book Progressive Delivery.
Does Platform Product Management & Design Really Happen? Or is it all just platform engineering? - Tanzu Catsup - Most organizations treat infrastructure as a series of projects to be “completed,” but successful platform engineering requires a permanent product mindset. In this episode, we explore why platform teams need dedicated product management.
You Can Feel It Coming - Software Defined Talk
New short videos since last time: bad advice from Wall Street on enterprise AI, devs want AI tools that freak ops people out.

Melany Lane typeface.

Logoff

I’m still figuring out how to setup my newsletter at micro.blog, and move off substack. It doesn’t work yet, so I’m posting this manually to the newsletter list (via substack, sure). But, this means, though, if you’re reading the newsletter that you missed the episode from earlier this week. You can check it out here if you like links and original content.

You Can Feel It Coming - Software Defined Talk

Fri, 20 Feb 2026 09:53:00 +0100

This week, we discuss personal AI hype cycles, bottoms-up adoption, and “The Modern Stack” simplifying cloud. Plus, thoughts on new cars and the dogs that ride in them.

See the traditional podcast listing for links and more.

Progressive Delivery, with Heidi Waterhouse - Software Defined Interviews

Fri, 20 Feb 2026 09:51:19 +0100

See the traditional podcast version for more and Heidi links.

Why it's great to be a Spring developer now, and how to make it even better - State of Spring, 2026

Fri, 20 Feb 2026 09:09:38 +0100

This is a talk I give at the start of Spring workshops we do. Here is the recording. The point is to show people that being a Java and Spring developers is fantastic right now. Here’s the description:

Spring developers are in a strange position in 2026: everything is changing: AI, platform engineering, enterprise architecture. And yet Spring keeps getting stronger.

In this talk, Coté walks through why this is actually a great moment to be a Spring developer, especially in large organizations. He looks at:

How AI code generation is changing Java and Spring development
Why enterprise momentum still matters
The role of platforms in making Spring teams faster
Where private AI, model brokering, and MCP-style patterns fit in
What you can do to increase your leverage as a Spring developer

If you’re building enterprise application, working on internal platforms, or wondering how AI changes your day job, this is a grounded, pragmatic look at where Spring fits in 2026, and how to push it further.

And here’s the slides.