Buying a used car is the largest purchase most people make outside a home, and on Syarah it happens without a showroom visit. A single order moves through inspection, financing approval, ownership transfer, and delivery — four processes, each on its own timeline, and a buyer who wants to know where all four stand.
The Alternative Was to Double the Team
Syarah set out to serve more customers, faster, at a point when the company was aiming to double its sales. The team went looking for a way to do that which did not route through hiring. As they describe it, the alternative on the table was simply to double the number of support staff — and they did not regard that as a practical answer.
It would also be the company’s first time building anything on AI. The worry going in was not whether the technology worked in general. It was narrower and harder to satisfy: whether an agent would actually understand a Saudi customer — the dialect, and the context behind the question — rather than matching words in it.
That concern is not abstract here. Syarah sells only in Saudi Arabia, its customers are Saudi, and around 95% of them would rather hold the conversation in Arabic. An agent that handles Arabic approximately is an agent that handles almost every conversation approximately.
Arabic Is What Separated the Vendors
Syarah worked through a long list of AI providers before committing to one. What decided it was not the demo. It was which agent held up in Arabic, because that is the language nearly every one of its customers writes in.
The second factor was that the engagement did not behave like buying a chatbot. Syarah’s team points to the working relationship itself as the thing that made it work — two teams building together, enough flexibility to change what was not right, and meetings that kept the same intensity after the contract was signed.
Building on the platform was also simpler than what they had seen elsewhere. The team had evaluated providers where standing up an agent was a complicated exercise; here it was not, and that is a large part of why they backed the partnership.
The rollout was deliberately gradual. Syarah started narrow, watched the quality of the conversations the agent was handling, and widened its scope each time the results held. Anything that genuinely needs a person still reaches one — the customer care team’s work became the cases that actually call for human judgment.
The Results
What changed once the agent was live, in the team’s own account:
| Metric | Impact |
|---|---|
| Response speed | Faster than hiring would have reached |
| Customer language | Arabic — about 95% of customers |
| Rollout | Gradual — widened each time quality held |
| Escalation path | Cases that need human judgment |
Response time is the number Syarah’s support organisation manages against, and it is the one the team reaches for first. Their own read is that the same result was not available by growing the team — that the speed they run at now is not a level more hiring would have brought them to.
The Impact
Faster replies changed how customers behave, not only how quickly they get served. Exchanges that used to stop after a limited number of replies between customer and agent now run further, and buyers ask the questions they actually have rather than the single one they were willing to wait for. The tickets themselves come out shorter.
For a company selling someone a car they will never see in a showroom, that engagement is the trust. Syarah’s team describes the outcome as targets they had set but had not expected to meet this well, and describes AI as now genuinely part of the customer journey rather than a layer sitting in front of it.
