Grab’s new AI tool makes menu management a snap for merchants

Since the feature was rolled out regionally three weeks ago, nearly 70,000 menu items have been added through the Merchant Menu Assistant’s scan option. It’s a sign that our GrabFood merchants are finding this feature useful and intuitive to use.

Solving problems on two fronts

Previously, setting up a menu on Grab would have required the merchant to type in each item manually. It was quite laborious, especially for restaurants with an extensive menu. 

Smaller, less tech-savvy merchants were sometimes overwhelmed with this task, and would ask Grab’s operations team for assistance during the onboarding process.

That’s why, initially, we tested the Merchant Menu Assistant with our operations team. They were already spending a significant chunk of their time helping small merchants with this task. The Merchant Menu Assistant made this much easier for them, and the operations team helped us refine the feature, until we were satisfied with its user experience flow and accuracy. 

Now, Merchant Menu Assistant is now available across the region for all merchants to use on a “self-serve” basis.

The tool automates a time-consuming task, which helps many of our food vendors, especially the smaller ones running their restaurants without a lot of manpower. A large number of the six million merchants and partners on Grab are small-to-medium F&B outfits, and they serve millions of customers throughout Southeast Asia. The quicker we can help them upload and update their menus, the faster they can start selling through the Grab platform and focus on running their business.

From hackathon to real-world implementation

Merchant Menu Assistant is a great example of how we innovate at Grab. The idea was developed at one of Grab’s Hackathons—these are internal competitions during which Grabbers are encouraged to think out of the box to build prototypes for new, ground-breaking features. An early version of Merchant Menu Assistant was the winning solution of Grab’s most recent Hackathon.

(Also read: How Grab’s tenth hackathon helped fast-track generative AI ideas from pilot to launch)

Technically, Merchant Menu Assistant works with a combination of optical character recognition (OCR) techniques, and Large Language Models (LLMs). The OCR helps capture text from images, and the LLMs ensure this text gets classified and organised into the right menu item data structure, such as menu titles, prices, and so on.

What’s next for Merchant Menu Assistant

We will continue improving this feature based on merchant feedback. One area we’re working on is being able to support all types of menu structures and complexities by exploring different OCR techniques. This includes ensuring high accuracy across all the languages we support in the region.

We’re also working on expanding the feature to scan more data.  Currently we only scan title, description, price, and category. But we are also looking at how to support modifiers—such as menu item options—and whether we can export dish images too. Beyond that, we’ll be looking at ways to make Merchant Menu Assistant a tool for not only scanning and uploading, but optimising merchant menus to drive sales. 

(Also read: How we use GenAI to help GrabFood menu items stand out)

AI technologies are a huge productivity booster in this arena, and we’re excited about the many ways in which we can make selling on Grab more delightful for our merchant-partners.