Lessons from the Ground 

AI in Retail

#LFTG | 9th July 2025

This LFTG, 25+ Operations and VM leaders from across Retail and F&B chains came together to share how they are already using AI or planning to use AI in the near term (1 year). They also shared the kind of impact they are sharing and the early learnings from the efforts. Here’s a glimpse into the kew takeaways:

How Retail and Restaurant chains are using AI

1. Visual Reviews With 200-500 pictures coming in from each store each month and no bandwidth to review them all, use of AI to review these pictures emerged as the single largest area where most participants have already started using AI. Others intend to.

2. New employee Onboarding A new employee joining in has hundreds of questions they need answers to. The use of AI as a buddy to enable them to get immediate and accurate answers without it becoming a burden to existing teams emerged as the second top area of AI use and impact.

3. Technical Knowledge Several complex products have loads of details. What’s the Tyre pressure for this bike model? What are the differences between these 2 device models? Not every sales person can remember every technical detail. Enabling the team with a tool, that can provide all details on call was the 3rd top area of AI impact in retail.

4. CCTV insights A restaurant or Grocery store receiving fresh produce has to follow a lot of steps: Receive the product, weigh it, enter details and finally inward it. The use of AI on top of CCTV footage was another area of use that emerged

5. New Store Opening AI based insights to feed into New store opening decisions was another area that emerged as a successful one as far as use of AI for Retail decision making was concerned.

AI generated MOMs, creatives and other Generative AI cases emerged as common areas of use across.

Top Benefits

Retailers using these AI technologies shared early Benefits in form of:

– Reduced effort | Feasibility Without AI, most teams would review some sample images from a few stores and share feedback. With AI, all pictures have started getting reviewed without any human effort and the reviews have become consistent. 

– Faster turnaround Manual reviews took time to complete (3-4 days for 500 stores) but with AI coming in, these have become immediate. With post review, corrected pictures coming in to the HO, this 3-4 days of review effort is completely eliminated.

– Increased Credibility An interesting benefit shared was that while stores would earlier upload random images, most of which would not show up in the sample checks, with all pictures being instantly reviewed by AI, random picture upload has completely stopped. At the same time, there was no impact reported on number of submissions with completion percentages remaining similar.

Additional benefits in form of sales impact, reduced costs/ manpower and more are planned, but most participants were in early stages of implementation and decisions that would eventually lead to such impact have not yet been taken. None of the participating retailers have yet started cutting jobs, put any hiring freeze or started replacing people with any form of AI.

The use of AI is currently focused on enabling existing people to do jobs they were so far unable to but can now do with AI and improving the quality of job done by them.

Top Learnings

Two key learnings that emerged from the discussion were as follows:

Plan for the fine tuning time While AI models come pre-trained, they do need fine tuning and there may be several edge cases that you may need to fine tune for. It can take anywhere between days to months to identify and fine tune for every edge case unless the provider has planned for such scenarios and has an approach for quick and easy fine tuning. One needs to evaluate service provider on their fine tuning capability.

Retail employees do not like to read long text AI models for learning / knowledge search often provide long text output/ analysis. And your employees are not going to read those long outputs. Go for crisp, to the point and easy to understand answers.

That’s all from this LFTG. But watch out for more Lessons from the Ground.