
Agentifying inbound support and debt collection for a leading Saudi fintech
A Saudi investment app moved support and early-stage collections to dialect-native Sarj voice agents, and cut peak call queues by 70%.
70%Shorter peak call queues

A Saudi motor insurer moved its first-notice-of-loss (FNOL) line to Sarj voice agents. Drivers can now report an accident in their own dialect at any hour, and 65% of FNOL calls are completed from first question to claim number without a handoff to an agent.
Reporting a claim is the moment a policyholder most needs a quick, calm answer. For this insurer, it was also the moment its contact center was least able to give one.
Accidents don't keep office hours. FNOL calls peaked in the evening and at weekends, when the claims team was thinnest, and callers who reached voicemail often called a competitor's tow service first.
Each FNOL call followed the same structure: verify the policy, capture what happened, where and when, collect the other party's details, and open the claim. Agents spent most of the call typing details into the claims system.
Callers were often stressed and speaking quickly in regional dialects, and the insurer's earlier IVR could not keep up. The insurer also needed every recording and record to stay in the Kingdom.
The agent identifies the caller and the policy, confirms everyone is safe, and routes any injury or emergency to a person immediately.
It then walks the caller through the incident, location and the other party's details in a natural conversation, and sends a link to upload photos of the damage before the call ends.
The claim is created in the insurer's system during the call, and the caller hears their claim number and next steps before hanging up. Complex cases transfer to a claims handler with the full transcript and the fields already filled in.
“Drivers get a claim number in a few minutes, at two in the morning, in their own dialect. Our handlers start each claim with the details already captured.”
FNOL is a short, structured conversation that matters a great deal to the person making it. Handing it to a voice agent that speaks the caller's dialect and acts in the claims system made the insurer available at the moment it counts, and gave its claims handlers back the time to work the claims.
The insurer is now extending the agents to policy renewals and claim status calls.

A Saudi investment app moved support and early-stage collections to dialect-native Sarj voice agents, and cut peak call queues by 70%.
70%Shorter peak call queues

A Saudi retail bank built Sarj voice and document capture into its app, so customers finish onboarding by speaking Arabic.
<120msIn-app voice latency

A Saudi clinic network reads, checks and files handwritten claim forms from more than 40 clinics with Sarj document agents.
80%Less manual data entry
Sarj AI is natively trained on regional Arabic datasets - it is not an English model translated on the fly. It understands localized accents and dialects (including Najdi, Hijazi, Gulf, Egyptian, and Levantine) as well as Modern Standard Arabic (MSA).

Sarj AI automates complex customer interactions, extracts critical document
intelligence, and executes backend operations end-to-end.