Meta Says AI Is Making It Easier to Build New Apps — and More Are Coming

Meta is using artificial intelligence to speed up app development, and the company says more new products are already on the way.

During Meta’s second-quarter earnings call this week, CEO Mark Zuckerberg said the company has additional apps in development following a recent wave of launches. Those have included a stand-alone app for Marketplace sellers, an app for Facebook Groups, a vibe-coded gaming app, a new photo app from Instagram and an experiment involving AI-generated bedtime stories.

Meta has spent years attempting to build successful stand-alone social apps alongside its core platforms, often without lasting success. Now, the company believes large language models can make software development faster, allowing teams to build and test new ideas more quickly.

“I’m … excited about how AI is helping our teams speed up product development,” Zuckerberg told investors Wednesday.

He pointed to recent launches including Instagram Instants, Forum, a stand-alone Groups app, and Seller, a stand-alone Marketplace app.

“I expect it to become a lot easier to ship new apps,” Zuckerberg said, adding that Meta plans to develop more ideas and use its recommendation systems to help them reach interested users.

Zuckerberg said AI is already improving Meta’s existing business by making its apps more relevant and delivering stronger results for advertisers, while also helping the company create new types of products.

Meta has tried the app-incubator strategy before

This is not Meta’s first attempt to rapidly experiment with new social applications.

In its earlier years, when the company was still called Facebook, it operated an internal incubator called Creative Labs designed to develop and test new social concepts.

Creative Labs produced several apps, including photo-sharing service Slingshot, anonymous chat app Rooms, Flipboard competitor Paper, photo-sharing app Moments and collaborative video app Riff.

The initiative ended in 2015, and the apps were eventually shut down after failing to build large, sustainable audiences.

Meta tried a similar strategy again in the early 2020s through its internal research and development group, the NPE Team.

That group experimented with a long list of products, including chat app Bump, social music app Aux, task app Move, dating app Spark, calling app CatchUp, zine creator E.gg, events app Venue, creator Q&A app Hotline, Cameo rival Super, couples app Tuned and music app BARS.

Once again, none developed into a breakout product, and the apps were eventually discontinued.

Threads gives Meta a more successful example.

This time, however, Meta can point to Threads as evidence that its newer approach to launching and scaling apps can work.

Threads has reached 500 million monthly active users, making it a significant success compared with many of Meta’s previous stand-alone app experiments.

Zuckerberg has repeatedly said he believes Threads could eventually become Meta’s next platform with more than one billion users.

One lesson from Threads was the importance of using Meta’s massive existing user base to seed a new service. The company promoted Threads extensively across Facebook and Instagram, helping it quickly attract users.

AI has also played an increasingly important role in the platform’s growth.

Meta says it has recorded “significant gains” from AI-powered content recommendations, which help determine what users see and keep them engaged.

LLMs are increasingly powering Meta’s recommendations.

Meta CFO Susan Li told investors that large language models are becoming increasingly effective at improving ranking and recommendation systems.

According to Li, LLMs help Meta’s existing systems better understand what content is actually about while also generating higher-quality training data.

LLM-powered agents are also assisting Meta’s engineers by evaluating content quality, identifying trends and testing changes to ranking systems.

Earlier this year, Meta reached another milestone in that effort. Every Reel and Feed post on Instagram is now automatically processed through an LLM, according to Li.

The system analyses characteristics such as the topic and tone of each post, giving Meta additional information that can be used to improve recommendations.

The company is also developing recommendation systems built natively around large language models, technology that could become increasingly important as Meta launches more stand-alone apps.

Better recommendation systems could allow the company to identify which users are most likely to be interested in a new product and promote those apps across Meta’s enormous existing network.

More Meta consumer apps are coming soon

Investors did not press Meta executives for additional details about the new apps under development during the earnings call.

Instead, much of the discussion focused on Meta’s rising AI spending and the company’s expanding ambitions in the enterprise AI market.

Zuckerberg nevertheless suggested that the next wave of products is not far away.

He said Meta has more “new consumer products” that will be “releasing soon,” signalling that the company intends to continue using AI to accelerate its renewed push into stand-alone apps.