I read the essay (https://unessays.substack.com/p/talk-is-cheap) and I started reflecting about the AI progression in my company for the last year.

I work at a small sized company with 25-30 developers. Since last year the company is enforcing AI on all levels, whether it be about writing tickets, coding, architecture, documentation, business decisions or user and market research.

As the essay suggests, we on a all-time low on shipping value to our customers or producing high-quality software. Our big rate is now about 26% of all our support requests, leaveling out the customers with quality regression. Our tickets are written with AI into Jira and consumed by another AI to produce something. Documentation is fully done by AI by some teams which habe a very high signal-to-noise ratio. My company has done a lot of research about our customers last year but the AI is unsure what to do. The company metric's are going up and done and nobody in charge has a novel idea what's going on as multiple LLMs are seeing different options. Our market research is done by AI without some insight and this is presented as 'we need to endure this and keep to the old plan'. Our product development is nearly stalled with 2 features shipped in about a year and some smaller iterations on existing features.

I'm currently pointing to the missing ideas and stalled value proposition for our customers but only earning 'we need more AI in our development'. Reviews and rework are our greatest bottleneck and nobody cares about. It seems to be seen as part of the progression instead of a guard rail to change something.

Do you have similar experiences when reflecting about the increased AI usage in your company? How do you takle the problem with junior developers which are now thinking they're seniors and above? What guard rails does you company use to ensure the value is reaching the customers?