Reengineering starts from a question most automation projects skip: should this process look the way it does? The answer is usually no. Processes accumulate steps that made sense once, a check added after an incident, a copy kept because two systems never talked, an approval nobody remembers requesting. Reengineering removes them before anything is automated.
Where did business process reengineering come from?
Two articles in 1990 started it. Michael Hammer's "Reengineering Work: Don't Automate, Obliterate" in the July to August Harvard Business Review argued that companies were using computers to speed up processes that should not exist. The same summer, Thomas Davenport and James Short published "The New Industrial Engineering" in MIT Sloan Management Review, arguing that information technology and process redesign belong together and describing a five-step approach drawn from nineteen companies.
Hammer and James Champy's 1993 book, Reengineering the Corporation, made the idea a management movement. IBM's current definition keeps the original intent: "the radical redesign of business processes to achieve dramatic improvements in performance, efficiency, and effectiveness."
How does reengineering work in practice?
The method is simple to state and hard to do:
- Name the outcome. A customer's service is active and billed correctly. A supplier is paid the right amount on time.
- Map the current process honestly, including the workarounds and the spreadsheets.
- Question every step. Who needs this? What would break without it? Does it exist because of a system limit that no longer applies?
- Design the target process with as few handoffs as the outcome allows.
- Choose the technology last, to fit the design.
Why did BPR lose its shine?
Reengineering programs in the 1990s were often run top down, at the scale of a whole company. IBM's overview of BPR notes that its "association with the downsizing and outsourcing trends that were unpopular at the time brought negative attention" to the method. The lesson Manthle takes from that history is to apply the thinking to one process at a time, with the people who do the work in the room.
What is BPR's lesson for AI?
Hammer's title applies again. An AI system built around today's process inherits its waste: the duplicate checks, the copy between systems, the approval that waits two days for a signature. The difference now is what the redesigned process can contain. A step that once needed a person only to read and interpret, such as reading a letter or matching a messy record, can be given to a model working within policy, while the judgment calls that need authority stay with a person.
That is why the Rethink step comes before the Engineer step in the AI implementation method Manthle uses, and why a business process automation engagement that starts by scripting the current steps has skipped the most valuable part. Process redesign is the smaller-scale form of the same discipline, and process mining supplies the honest map of the process as it runs today.
