This has been one of our main tenets with NServiceBus and the rest of the platform in Particular. Everything should always Just Work™. https://t.co/FbGscZ26GS
Moved a running system to new servers with no downtime today thanks to @ParticularSW NServiceBus. Also love patching parts of the system with no downtime! #24x7#uptime
Another team build a customized framework that does everything (see Inner Platform Effect.) Today I have to meet with them and they will pressure us to start using their platform instead of just writing our own code like our forefathers before us. 1/
Every IBM MQ project reinvents the same plumbing.
MQGET loops. Backout logic. Connection management. Every project writes the same infrastructure code, just slightly differently each time.
We just released an IBM MQ transport for NServiceBus. The split is clean: IBM MQ moves bytes, NServiceBus handles what your application code does with them. The concept mapping is direct: your MQGET loop becomes a handler, BackoutCount becomes automatic retries, and syncpoint becomes atomic receive + send.
Pub/sub is where it gets interesting. You define a C# type, call Publish, and write a handler. NServiceBus creates the topic objects and durable subscriptions in IBM MQ behind the scenes. No manual runmqsc scripts for the common case.
So atomically committing a database write and a message send isn't natively possible. NServiceBus closes that with the Outbox - outgoing messages go into the same database transaction as your business data. Exactly-once processing, no distributed transactions needed.
https://t.co/uEbX7l6xFq
@NoVibeJustOrch@brankopetric00 Could you explain more about what "an NServiceBus failed" means?
Because, as stated, that would essentially be equivalent to "an https://t.co/YXAQ3cjMJm failed".
If you're interested in the idea of vertical slices and ViewModel composition, you should check out ServiceComposer, which now includes OpenTelemetry support! https://t.co/QG2AfUhuEq
I recently came across a team using an LLM to implement a state machine.
They called it "agentic". They also told me it worked correctly most of the time.
I asked them to write down the list of rules for moving from one state to another.
They listed them out.
It didn't take long for them to realize what was happening: They had built a solution using AI for a problem that didn't need AI.
We removed the model and wrote some code to implement the rules. It might have taken 2-3 hours at most.
They went from "it works most of the time, it's relatively fast, and it costs some tokens" to "it works 100% of the time, instantaneous, and costs nothing."
So many examples like this recently.
Golden rule: BUILD THE SIMPLEST THING THAT COULD POSSIBLY WORK.
"And now they've left"
Features missing/incomplete. No docs or tests.
And bugs… oh so many bugs.
These are the stories of the devs left dealing with custom frameworks after those who wrote them leave.
Don't do that to your colleagues.
As much as possible, use existing tools.
More and more devs noticing the "how hard could it be" trap.
If you're in London, go chat with Luca.
In any case, join the community forming around this:
https://t.co/bZKOnh4kmP
There's a moment that happens in every engineering team I've worked with, and I don't think we've named it yet.
It's the moment someone says "how hard could it be?”
Don’t get me wrong, they mean it sincerely.
I've watched smart, experienced people say it about a date picker or a search bar, even about a permissions model.
Each time, the same arc: a few weeks of confident progress, a long middle section of strange edge cases, and eventually a quiet realisation that the problem was never the thing they thought it was.
Before #AI, this happened maybe once or twice in a career and most teams end up buying the component.
Now it happens way more often because AI doesn't just generate the first version, provides the confidence people are looking for.
The code looks right, all tests pass and even the demo works.
On 19th May I'm filming a documentary at "Beyond the Prompt" organised by @ag_grid in London on exactly this question: in 2026, what is genuinely worth building yourself, and what are we going to regret?
If you've lived through a "how hard could it be?" that turned out to be very hard indeed, I want to hear it on camera!
Tickets are still available 👉 https://t.co/8DSWVxSAZE
Plus, I got a gift for YOU! 😉 read the comments...
If AI makes building and maintaining software much cheaper: Amazon is making massive mistake in retiring their own response to Zoom (it should be v cheap to maintain!)
If maintaining software remains expensive even with AI, rebuilding a nontrivial SaaS in-house could be a mistake
I do appreciate the enthusiasm to rebuild stuff, and we’ll learn in 1-2 years if it was worth it! Trying better than speculating tbh
Source of Amazon announcement: https://t.co/tMJj87Is5W
@TrungTPhan Those stats are likely from 2017.
https://t.co/TniAMa5FFs
> A 2017 Reuters study lists 43% of banking systems still use COBOL
Probable source:
https://t.co/Q3cBmyfsZ1
Do you have anything more current?
Custom in-house software frameworks are like buying too many apples. You start telling people they have to eat the apples before they go bad.
It stops being about what people need and it becomes about what the apples need.
Another, really well articulated case for "companies shouldn't vibe code all their 3rd party core apps".
Yes, it may save some dollars, but it's a major distraction.
If you have the kind of talent that can leverage AI to solve problems, work on things that are core to your mission and serve customers.
Spend every possible calorie you have on creating customer value and moving the business forward.
A couple of months ago, I was speaking at a conference when it came time for the Q&A. Several people raised their hands, and one question after another came out. Interestingly, every question that was asked was something I had already written about in a blog post, sometimes many years earlier. Because I had already put my thoughts on paper, my answers were more thoughtful, more structured, and more relevant.
After that experience, it made me reflect on something that feels obvious but is rarely said: the more you write, the better you speak.
Writing forces you to slow down and think. When you write an idea, you have to work through the pros and cons, document the anecdotes, and clarify the lessons. That effort crystallizes your thinking and refines your word choice. Writing is one of the best mental exercises for improving how you think, which directly improves how you speak.
There is also a powerful bonus. Writing captures your ideas for posterity and helps you retain them far longer than simply thinking or hearing them.
My advice to entrepreneurs is simple: start writing. It does not have to be perfect. It does not have to be groundbreaking. It just has to be done. The more you write, the sharper your thinking becomes, and the stronger your speaking becomes. Writing and speaking reinforce each other, and entrepreneurs should invest in both.
Mainstream support for NServiceBus version 8 will end on April 16, 2026. Extended support plans are available. Check out all the details here: https://t.co/kmJHP11lCO
@codeopinion If you're not using some kind of framework like NServiceBus or MassTransit, it can be hard to test more complex message-driven logic.
With those kinds of frameworks, no biggie:
https://t.co/krHC4s94lO