# AI should upgrade the job, not cut it

> The boss of 17,000 health and beauty stores is measuring AI by customer love, not headcount saved.
- Title: AI should upgrade the job, not cut it · Fortune | Shortify
- Summary: The boss of 17,000 health and beauty stores is measuring AI by customer love, not headcount saved. If you point AI at today's processes, you just get a…
- Keywords: retail, jobs, automation, leadership, upgrade, job, cut, Fortune
- Source: Fortune — https://fortune.com/2026/09/12/as-watson-ceo-malina-ngai-ai-retail-jobs
- Author: Nicholas Gordon
- Read time: 1 min
- Topics: ai, retail, jobs, automation, leadership
## The question leaders skip
AS Watson Group CEO Malina Ngai put it to the Fortune Leaders Forum on Sept 8: the choice is not obvious, and most firms pick the cheap answer.

> Is it really that, with AI, we're going to cut out 30% of our workforce? Or… are we going to increase the quality of our people's intelligence?
## The industry went the other way
Tech firms have cut tens of thousands of roles this year while pivoting to AI. Oracle alone shed 21,000 jobs, about 13% of its workforce.
## Bring the assistant to work
Ngai's alternative is a human-AI partnership: every employee works alongside an assistant that speeds up their thinking rather than takes their seat.

> You should bring the AI assistant with you to work. It helps you to think faster and smarter.
## Measured by customer love
Since launch last October, engagement scores jumped and a "customer love score" keeps rising as staff spend time with people instead of screens.

> This is not the usual way that people would measure AI.
## Robots changed who gets hired
A warehouse in Foshan is normally 80% men because of the lifting. With robots doing it, 62% of that site's workers are now women.

> That was something that really positively surprised me, how technology can also bring more diversity in the company.
## A version of yesterday
Ngai's warning: aim AI at existing processes and you automate the past. The gain comes from rethinking the model, which is why she expects retail to get more human.

> Basically, you get a version of yesterday.
## Key takeaway

If you point AI at today's processes, you just get a faster version of yesterday. Point it at the assumptions instead.