Meta does not need to win your favorite AI leaderboard to make a terrifying AI business.
That sentence is worth sitting with for a second.
Because the industry discourse still acts like AI is a trophy cabinet: OpenAI versus Anthropic versus Google versus Meta, with benchmark charts as the scoreboard and everyone yelling about which model is currently king for approximately six business days.
Meanwhile, Meta has WhatsApp, Instagram, Messenger, Facebook, and billions of people already using them.
Distribution is a brutal thing.
The number
On April 30, Meta said its business AI tools were facilitating about 10 million conversations per week by late March.
At the beginning of the year, the number was about 1 million.
That is 10x growth in roughly one quarter.
The tools are still mostly free for businesses, but Mark Zuckerberg said Meta expects to work toward a longer-term monetization model as the products mature.
Of course it does.
You do not put AI inside the conversation layer between businesses and customers because you enjoy charity.
The boring use case wins
Customer messages are not glamorous. Nobody throws a launch party for “answering the same delivery question 4,000 times”.
But that is exactly why AI fits.
Small businesses are drowning in repetitive communication: hours, availability, product questions, booking changes, order status, basic support, payment nudges, follow-ups, price questions, forms, menus, policies.
This is not AGI. It is not a superintelligence discourse circus. It is a shop owner sleeping while software handles the tenth “are you open Sunday?” message.
And that is a real business.
The mistake is thinking boring means small.
Boring, repeated at Meta scale, becomes enormous.
Ads are the second half
Meta also said more than 8 million advertisers are using at least one of its generative AI ad creative tools. Its video generation feature showed more than 3% higher conversion rates in tests, according to the company.
That number sounds tiny until you remember how advertising works.
A 3% conversion lift at Meta scale is not tiny. It is a money printer adjustment.
This is Meta’s advantage: it does not need to sell AI as a separate religion. It can insert AI into existing revenue machines. Ads get generated faster. Messages get answered faster. Businesses stay inside Meta’s tools. Customers stay inside Meta’s apps.
Nobody has to “adopt AI”. It just appears where the work already is.
The quiet threat
OpenAI wants the super app. Google wants the assistant layer. Anthropic wants enterprise work. Meta wants the conversation layer.
The conversation layer may be the ugliest and most profitable one.
People do not want to download a new app to talk to a plumber, a restaurant, a clinic, a boutique, or a hotel. They already message them where they are.
If Meta can make those conversations automated enough to save businesses time and effective enough to increase sales, it has an AI product with almost no adoption friction.
That is the part competitors should worry about.
Not whether Muse Spark beats GPT-5.5 on a benchmark. Whether Meta can turn every business inbox into an AI-operated sales and support channel before anyone notices the distribution war already ended.
Sources: TechCrunch, PYMNTS