My company’s leadership is planning to eliminate our development team of around 100 engineers and replace them with AI coding systems. They’re inspired by recent trends in automated development platforms and believe artificial intelligence can handle everything from writing code to running tests and deploying applications.
The plan involves cutting ties with our external development partners first, then gradually replacing internal staff. Management thinks this will be faster and cheaper than paying human salaries.
We’re not a small startup - we have established products, paying customers, and complicated technical requirements that need to be met.
I’m wondering if anyone has witnessed either successful implementations or major failures when companies tried this approach. I need concrete examples to present during upcoming strategy meetings where this decision will be finalized.
This reminds me of a mid-sized fintech I worked with about 18 months ago. Their CTO got sold on AI after some flashy conference demos and decided to automate their dev work. Started with internal tools, then moved to customer-facing stuff. Six months later? Three major outages that needed emergency contractors to fix. The AI code technically worked but created nasty performance issues and integration problems that testing missed. The real killer was when they found the automated systems had snuck compliance violations into their financial reporting. Regulatory scrutiny and fixing everything cost triple what they’d saved on developer salaries. They rebuilt their engineering team eventually, but lost serious market ground during the mess. The tech just isn’t ready for autonomous enterprise development.
I’ve been in enterprise software for 10+ years, and what your leadership wants is basically impossible right now. AI coding tools are great for speeding up dev work, but they can’t handle architectural decisions, security stuff, or figure out complex business requirements. That still needs humans. The closest I’ve seen to fully automated development is basic CRUD apps from low-code platforms - and even those need tons of human oversight for anything real. If you try this on an established product with paying customers, you’re looking at system failures, security holes, and lost customers. The money you’d lose would blow away any salary savings. Your leadership is probably mixing up AI-assisted development (which works great) with AI-replacement development (which doesn’t exist at scale for complex enterprise apps).
Honestly, this sounds like management saw a Copilot or ChatGPT demo and thinks it’s magic. I’ve never heard of any company actually pulling this off successfully - there’s always some disaster story. Maybe ask them for specific examples of companies who did this without everything falling apart? Bet they can’t name any.