Meta Business Agent Gets Stilla.ai's Team to Turn Chats Into Sales
Meta has acquired Stilla.ai, a Stockholm startup whose agents turn customer conversations into completed purchases, and will fold its team and technology into the Meta Business Agent that operates across WhatsApp, Messenger and Instagram. The company did not disclose financial terms. Stilla was founded in 2024 by Siavash Ghorbani and Kaj Drobin, both former Shopify executives, and emerged from stealth earlier in 2026, which gave it roughly eight months as a publicly launched company before the sale.
The deal adds a small engineering and product team to Meta's business messaging organisation, the unit that owns commerce across its three chat apps. Both founders previously built the Swedish e-commerce platform Tictail, so what Meta is buying has a commerce-operations pedigree rather than a pure research one.
What the Meta Business Agent Already Does
The agent itself is not new. Meta began selling businesses access to an AI assistant in June 2026, the first time it charged merchants directly for software that could act on their behalf instead of merely replying. The product schedules appointments, completes transactions, resolves routine customer issues, recommends products and passes complex cases to human staff.
A companion layer, the Meta Business Agent Platform, lets merchants connect third-party data from services such as Shopify and Zendesk, so the assistant can personalise answers using order history and support tickets. That integration surface is part of the logic behind buying Stilla: the startup's founders came out of the Shopify ecosystem, the same one Meta's platform is built to plug into.
Meta presented the June launch as a way to build revenue beyond advertising. Selling agents to businesses is a direct attempt to create a second income line that can offset the cost of the company's AI infrastructure, and Meta has signalled it will raise planned AI investment again for 2026.
From Answering Messages to Closing Sales
The shift inside business messaging shows up more clearly in the contrast than in the announcement itself.
| Dimension | Ad-led messaging | Agent-led messaging |
|---|---|---|
| What Meta sells | Impressions and click-to-chat placements | A hosted assistant that acts for the merchant |
| Who pays | Advertiser, per campaign | Merchant, for agent access |
| Who owns the conversation | The merchant | Shared with Meta's software |
| Cost of failure | Wasted ad spend | Refund, chargeback, lost trust |
| Integration | Link out to a chat thread | Connectors to Shopify, Zendesk |
For most of the past decade, Meta monetised chat through advertising. A merchant bought a click that opened a WhatsApp thread, and everything after that was the merchant's problem. The Business Agent inverts the arrangement: Meta now sells the conversation, the resolution and, increasingly, the checkout.
That is a harder product to ship than a chatbot. A transaction requires the agent to read inventory, apply pricing, take payment and accept the consequences if any step fails. Answering a question badly is an annoyance. Closing a sale badly is a refund, a chargeback and a merchant who stops trusting the channel. The capabilities Meta lists for the agent, from completing transactions to escalating complex cases, describe the exact seams where an automation layer has to be dependable.
Stilla's own framing pointed at that seam. It built an agent designed to retain a business's context and act on it, which is the unglamorous plumbing that decides whether an assistant can be trusted with real customer intent.
An Eight-Month Sprint to an Exit
Stilla's path is unusually compressed. Founded in 2024 and flagged as an early-stage Swedish prospect in December 2024, it only left stealth in 2026, and it now sells into a company with billions of users. That timeline says as much about the market for agent startups as it does about Stilla. Buyers are paying for teams that have already solved the messy parts of agent deployment, and they are paying early, before a startup's revenue makes valuation arguments complicated.
The company had attracted venture backing and was treated as one of Sweden's most closely watched young firms before the deal. An exit this fast usually signals that the founding team judged distribution, not independence, as the binding constraint. Building a merchant agent inside Meta's messaging apps removes the hardest problem any standalone agent vendor faces: getting in front of businesses that already run their customer conversations somewhere.
Why Meta Bought Rather Than Built
Meta runs one of the largest AI research organisations in the world, so buying an eight-month-old company with a handful of staff is itself the signal. Model capability was not the constraint. The constraint is productised merchant workflow: the specific logic that turns a language model into something a small retailer will let speak to paying customers.
The contrast with Meta's own roadmap is instructive. Its June work centred on connectors to Shopify and Zendesk, which is infrastructure. Stilla brings people who built commerce tooling at that layer, and dropping a small team into an existing organisation is usually faster than staffing a new one from scratch. The trade-off is dependency. Meta inherits Stilla's integration assumptions and its technical debt, and it paid an undisclosed price for a company with no public revenue figures.
Set against Meta's AI budget, the purchase is small. The company is raising its planned AI investment for 2026, and a team-sized acquisition is a rounding error next to data centre and compute commitments. The significance is not the price but the direction: capital is flowing toward the layer where AI touches a customer's wallet.
The competitive stakes run in several directions. Merchants get a shorter path from a message to a completed order, which matters most for small sellers who cannot staff a support desk around the clock. Platform vendors including Zendesk and Shopify face a partner that is increasingly also a rival, since Meta's agent can absorb work those tools currently charge for. The enterprise agent market Meta entered in June is crowded, with every major model provider and business software vendor selling some version of the same promise.
Buying the team also keeps a young competitor off the board. Stilla had eight months of merchant-facing work behind it and a founders' track record in e-commerce software, which is precisely the profile a well-funded rival would have wanted to acquire instead.
One cost no acquisition removes is trust. Letting an agent take payment and confirm orders means merchants are handing over the part of the sale that decides whether money moves. Escalation to human staff is the fallback, but the economics of an agent only work if escalation stays rare.
What to Watch
Two signals will show whether the Stilla deal pays off. The first is how quickly Meta folds the team's capabilities into the Meta Business Agent's transaction path rather than parking them in a separate product. The second is whether merchant adoption reaches a scale that makes the agent a material line against advertising revenue. The product is being positioned around a base of one million businesses, far beyond early adopters.
Neither signal is visible from outside Meta today, and both will take quarters rather than weeks to read. For buyers, the practical question is narrower. Any business already handling customer conversations on WhatsApp, Messenger or Instagram should test whether the agent can close a real order end to end before routing more traffic through it. The acquisition does not change the product today. It changes how quickly the product is likely to change.
Why this matters
Meta's purchase of Stilla.ai is a small deal with a large directional signal: business messaging is moving from a channel Meta rents to advertisers into a service it sells to merchants. If agentic checkout works at scale, the messaging apps stop being a place where conversations happen and become a place where money changes hands, which puts Meta in direct competition with the commerce and support software it currently integrates with. Merchants gain a faster route to a sale and a new dependency on Meta's agent to deliver it.
Related Articles
- Meta Launches AI Business Agent to Diversify Revenue Streams
- Meta Launches Facebook Seller App for Marketplace Power Sellers
- Meta AI agent stall: Zuckerberg admits four-month lag
✔Human Verified
Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.