I have a SAFe certification.
I’m not proud of it. I’m not ashamed of it either. It’s just a fact, like a scar from a surgery you barely remember. I sat through the training, I passed the exam, I got the badge. And then I watched as the thing that SAFe was supposed to represent - iterative, adaptive software development - got buried under a mountain of certifications, consultants, and two-day workshops that taught people how to move sticky notes on a board.
Agile started as a rebellion. A group of developers said: “We’re tired of building the wrong thing for two years. Let’s talk to users and ship in small pieces.” That was it. That was the whole revolution.
And then the machine showed up.
The machine doesn’t care about ideas. It cares about content, courses, conferences, and consulting hours. It took “let’s ship in small pieces” and turned it into a $20 billion industry of frameworks, role definitions, and maturity assessments. By the time most teams encountered “Agile,” it wasn’t a philosophy anymore. It was a product.
The developers who could have benefited the most - the ones drowning in waterfall timelines and spec documents - looked at what Agile had become and said: “That’s not for me.” And honestly? They were right. What they were looking at wasn’t Agile. It was Agile™.
I know this because I was there. I lived it. I bought the certification. And eventually, I ignored most of it and just focused on shipping.
The Pattern Has a Name
Here’s the thing nobody told me when I was sitting in that SAFe classroom: this wasn’t an Agile problem. It was a pattern.
Every genuinely useful idea in technology goes through the same cycle:
- Practitioners discover something that works.
- They share it, honestly, with caveats and context.
- A content industry forms around it.
- The content industry strips away the caveats and context because nuance doesn’t go viral.
- The idea becomes a brand.
- The brand attracts people who sell the brand, not the idea.
- The practitioners who started it all look around and don’t recognize what they built.
- Smart people on the outside see the brand, not the idea, and walk away.
Agile went through every single step. Microservices went through it. DevOps went through it. And right now, AI is somewhere around step 5, accelerating hard toward step 6.
Open LinkedIn. Close LinkedIn.
I open LinkedIn. Here’s what I see:
A post about how AI will replace 80% of software engineers by 2027. Twelve thousand likes. The author is a “transformation advisor” whose last technical contribution was a WordPress site in 2014.
A carousel explaining “10 prompts that will 10x your productivity.” The prompts are things like “be specific” and “provide context.” Groundbreaking.
A thread about how someone “built a SaaS in 48 hours with AI.” No mention of testing, deployment, error handling, or what happens when a real user touches it. Just a demo video and a signup page.
A thought leader announcing they’re now offering “AI Strategy Consulting for Engineering Leaders.” Their qualifications: they read the OpenAI blog.
I close LinkedIn.
I open my terminal. I look at the AI tools I’ve been building with - the ones that fetch requirements from a ticket, build context from a codebase, access logs to understand what went wrong, generate a fix, run the tests, read the failures, and iterate until it actually works. The loop isn’t magic. It’s engineering with a faster feedback cycle. And I think: we are not talking about the same thing.
The gap between the AI discourse and the AI reality is so wide that they might as well be different subjects. One is about magic. The other is about plumbing. Embeddings, latency, cost, hallucination rates, evaluation, and the boring, unglamorous work of making something reliable enough that you can go to sleep without worrying about it.
The Co-Villains
It would be easy to blame the influencers. Too easy.
The truth is, the echo chamber has co-villains, and I’m one of them.
Every time I scroll past a shallow AI take and feel the urge to “well, actually” in the comments - that’s me feeding it. Every time I share a post not because it’s good but because I want to be seen as someone who’s “in the conversation” - that’s me feeding it too. And when I stay silent because engaging with the noise feels beneath me? That’s me letting it win by default.
The influencers are the most visible carriers, but they’re not the only ones. The consultants who repackage blog posts into “AI transformation roadmaps” are carriers. The conference organizers who book speakers based on follower count, not shipping history, are carriers. The engineers who dismiss AI entirely because the discourse is embarrassing are carriers too - they’re ceding the conversation to the loudest voices.
The pattern doesn’t need villains. It just needs participants. And we’re all participating, one way or another.
The Part Nobody Says Out Loud
Here’s what I haven’t seen anyone write about, and it’s the thing that bothers me the most.
I feel lonely.
Not in some dramatic, existential way. In the specific, frustrating way of someone who loves a thing and watches it get misrepresented daily.
I’m genuinely enthusiastic about AI. I use it every day. I build with it. I’ve seen it do things that would have taken me weeks, done in hours. I’m not a skeptic and I’m not a doomer. I’m a builder who thinks this technology is one of the most important shifts in how we make software.
And I open my feed and I don’t recognize the thing I work with.
The AI in the discourse is a magic wand that replaces developers and writes perfect code. The AI in my terminal is a powerful, flawed, sometimes brilliant, often frustrating tool that requires real engineering judgment to use well. I don’t even know what to call the gap between those two things. But almost nobody is talking about the second one.
That’s where the loneliness comes from. Not from being contrarian, but from feeling like the real conversation isn’t happening.
And alongside the loneliness, there’s disappointment. Because AI deserves a better discourse. The technology is genuinely powerful. The things you can build with it are genuinely new. But the conversation around it has been colonized by the same machine that colonized Agile, and the signal-to-noise ratio is collapsing.
And then there’s the anxiety. The quiet, irrational hum that says: what if I’m wrong? What if everyone else sees something I don’t? What if the people posting the confident takes actually know more than I do?
I know that’s not true. I know it the way I knew the Agile consultants didn’t have some secret knowledge I lacked. But the feeling doesn’t care about what you know. The feeling cares about volume, and the echo chamber has a lot of volume.
If you’ve built something real with AI and felt any of this - the loneliness, the disappointment, the low hum of doubt - I want you to know: it’s not just you.
What We Lost With Agile
Let me tell you what the Agile echo chamber actually cost us, because I think we’ve forgotten.
It didn’t kill Agile. Agile still works. Small teams shipping iteratively, talking to users, adjusting course - that never stopped being a good idea. What the echo chamber killed was adoption by the people who needed it most.
The best engineers I know, the ones with the deepest judgment, the ones who’ve been burned by bad processes - they heard “Agile” and rolled their eyes. They associated it with stand-up meetings that accomplished nothing, retrospectives that changed nothing, and certifications that proved nothing. They opted out.
And because the best people opted out, the implementation of Agile in most organizations was led by the people who bought the brand, not the ones who understood the idea. That’s how you end up with “Agile transformations” that somehow make teams slower. I’ve seen it happen more than once.
The window where Agile could have been adopted thoughtfully, by serious practitioners, with nuance and context - that window was consumed by the machine.
AI has the same window open right now. And I can see it closing.
The Window
The engineers who should be experimenting with AI, the senior ICs with deep domain knowledge, the tech leads who know where the real bottlenecks are, the managers who understand what their teams actually need - many of them are tuning out. Not because they’re luddites. Because the discourse is unbearable.
When every other post is “AI will replace you” or “I 10x’d my productivity with one prompt,” the rational response for a serious person is to close the tab. And that’s exactly what’s happening.
The irony is painful. The people most capable of using AI well are being repelled by the people talking about it the most. Just like Agile. Just like every time.
The Antidote
I don’t have a framework. I don’t have a five-step plan. I’m not going to sell you a course.
What I have is something I learned the hard way, sitting in that SAFe classroom, watching the pattern play out in real time and not having the words for it yet.
The echo chamber doesn’t die from criticism. It feeds on it. Every “hot take about hot takes” is just another post in the feed. Every “LinkedIn is broken” thread gets the same engagement as the thing it’s complaining about. Fighting the echo chamber is the echo chamber.
The only thing that ever broke through the Agile noise was people quietly building good teams and sharing what actually worked. Not frameworks. Not manifestos. Just: here’s what we did, here’s what happened, here’s what we learned.
I think the same is true for AI. I hope so, at least.
Build something. Not a demo, not a weekend hack you’ll never touch again. Something real. Something that breaks and needs fixing. Something that teaches you the difference between what AI promises and what it delivers.
Then write about that. Not your opinion about AI. Not your take on someone else’s take. Write about what you built. What surprised you. What failed. What you’d do differently.
The next time you feel the pull to engage with the echo chamber - the FOMO, the pressure to have a take, the anxiety that maybe you’re falling behind - close the tab. Open your editor.
The discourse will sort itself out the same way it always does: when the builders outnumber the performers.
Be a builder. The echo chamber has enough voices. It doesn’t need yours.
It needs your demo.