All episodes
Legacy is an ownership problem
If you are trying to get a legacy system retired and cannot get it prioritised, we go through what stalls that work, how to make the saving land with a business, and the sequence we use to finish one.
Chapters
In this episode
- Where Sepehr Lorestani draws the line between old code and code nobody owns
- Fixing forward as the new standard, and what changes when an agent bumps the version
- Whether a decommissioning stalls on technical difficulty or on priority
- Finding a champion with decision-maker power, borrowed from sales
- Why a saved cost reads smaller to a business than a sale of the same size
- What one migration off a data centre to serverless was estimated to save
- How Fernando Gonçalves shrank a legacy system without stopping the business
- Sepehr Lorestani’s six-point list for finishing a decommissioning
- Whether Anthropic is a useful example for anybody else
- AI for reverse engineering a system nobody owns, against AI writing new code for it
- Whether every system should be designed for its own sunset


