Micromanagement in Oracle tied to AI surge

[Jasmin]

Background about me: I'm the first one in my team to use any of the 'lunch' and learn sessions tools. I also train others and get a lot of exposure and token usage. I've delivered projects in about 60% of the time thanks to AI.

Managers are handing out AI tasks, directors stay quiet and cautious. Both are fine.

The real problem is a bunch of leads and architects keeping saying things like:

"Why not use this AI tool?"
"Don't include that in the prompt."
"Did you ask AI?"
"Hey, have you finished the code (AI‑generated)?"
"Team X says this tool is ten times better, install it, use it."
"Team X did it in a few seconds with AI, why is it taking us weeks?"

I'm literally getting grilled for writing even 5% of the total code, while the design credit goes to AI even though it has no idea and just spits out code based on the design I fed it, which totally fried my brain.

AI can spin up a full backend in minutes, but I still have to review it, give feedback, prompt again, review, test each time.

Then I get messages like, "Hey, did AI generate the code? Are you done?" For a complex enhancement that would normally take 40–50 days, they are asking for updates every 3–4 hours.

Man, it's frustrating and it feels really demotivating.

[Raj]

Show them a 1500‑line markdown file generated by AI and tell them that's the design, ask for comments. I know they'll just use AI again to summarise it.

Markdown is easy; the real challenge is getting AI to write well‑structured code from scratch.

[Anand]

True, they have stopped valuing and respecting us like they used to before AI.

[Pratik]

I just overheard someone say:

"Non‑technical people are cranking out software so fast. We, the technical team, aren't doing that."

My take: that's exactly what will happen. Non‑technical folks don't have to worry about quality. If it works, it works. No one is going to stress‑test their little proof‑of‑concept. But for us developers, the code has to be production‑ready, scalable, secure and maintainable.

Non‑technical users just want something that works right away, even if the code is a mess. They don't need anyone to read it; they just ask the LLM and get annoyed when it fails.

[Aftab]

I built an entire Twitter clone with Cursor in just 15 minutes.

Check this out index.html

The UI even looks better than the original Twitter. It's over for you developers.

[Aravind]

Bro, it's not just Oracle. Any fast‑paced, wannabe company is doing the same, including my own.

[Advik]

I hope once all this AI‑generated junk reaches production and starts causing security holes and bug tickets, they realise their mistake. We don't know how long that cycle will last or whether we'll still be around to see it—maybe we'll burn out or get replaced by someone churning out all the junk they want.

For now the safest bet seems to be to play by their rules: spend tokens, focus on cranking out more code rather than quality, because that's where the incentives lie right now.

By the way, if you couldn't tell from my earlier comment, I'm frustrated with this trend too

[Govind]

Leads and architects?

Let's guess—they're from service firms, maybe without a CS degree, just a lot of 'matkas'.

[Mohit]

Don't worry, this 'AI' will only replace the toxic managers, not the developers. A flat hierarchy is the future.

Can an LLM write production‑ready code? No.

But can it generate PPTs, Excel sheets, resource plans, etc.? Absolutely.

[Naresh]

I felt the same today after a grilling meeting to cut down my estimates. Things are changing so fast that we're constantly expected to use the newest tool released that morning.

I tried explaining that code can be generated in a day, but it still needs review, validation, debugging, and further prompting to fix issues. I have to make many decisions, analyse designs and weigh trade‑offs of different approaches. I even noted down every time AI wasn't helpful or sent me on a wild goose chase.

But nope, on Monday I have to try the new shiny thing and report back by Tuesday.

And the best part:
If it works, AI gets the credit as usual.
If it doesn't, it's on me and my inability to prompt it better.

[Pratik]

Every corporation now feels like a fast‑paced startup.