Any AI SDLC workflow which involves taking a spec and turning it...

3 views Nov 15, 2026 ~1 min read

                            1
                        
Any AI SDLC workflow which involves taking a spec and turning it into a backlog of issues or tickets is a dead end. 

There is only the spec. Any materialization of the spec into discrete units of work can only introduce drift. It is waste.

                            2
                        
The problem with ticket-only workflows is there is no homeostatic point. If you give an agent the power to introduce new tickets, there is no back pressure on them to stop. You know what the back pressure is? Spec. When spec == impl, homeostasis.

                            3
                        
What is the point of a ticket? It’s a discrete unit of a spec, a deviation from the platonic ideal of the system. They only existed because of coordination problems when human SDLCs took months to close them. That’s all going away. It takes minutes to close tickets now.

Actions

What You Can Do

Export as PDF or Markdown
Batch Export to Notion
Bookmark & Highlight
Screenshot Tweet

Create Free Account

Includes 7-day Premium trial

Actions

What You Can Do

More in AI & Machine Learning