Webinar Recap: How Nana Transformed Customer Experience with AI Agents
Key point
Nana cut customer support costs by 85% and reduced response times to seconds by adopting AI agents.
Details
Nana, a major super app in Saudi Arabia, moved away from scaling customer support purely by adding headcount and put AI agents at the front line instead. Previously, 54 support staff worked in shifts to handle demand, but now agents inside the app directly handle customer conversations, taking care of responding to inquiries, tracking orders, resolving issues, and escalating when necessary.
The results were immediate. Customer support costs dropped by 85%, and response times fell to seconds. Existing support staff were freed from frontline responses and redirected to improving AI workflows, or to upgrading operations in other departments based on their understanding of customer complaints and operational issues.
During adoption, Nana tested several CRMs and AI vendors but didn't get the level of automation they expected. Some were effectively repackaging ElevenLabs technology and selling it at a higher price, so Nana ultimately chose to connect directly to ElevenLabs. As a result, a chat agent went live in actual production, and a voice agent for outbound calls is also operating and being expanded.
Internal operations saw major changes as well. Projects that would have taken weeks with traditional methods were built in a day and connected to backend systems. When new operational challenges arise, the default question is no longer how many more people to hire, but how quickly an agent can be built to solve it.
Here are the practical tips Nana shared:
- Design for multimodality so customers can interact via text as well as phone calls
- Invest in prompt quality rather than the technology itself, so agents follow the brand's tone and express empathy
- For Arabic deployments, add tashkeel to improve pronunciation accuracy and naturalness
- Instead of reducing remaining staff, move them to higher-level work such as designing AI workflows and quality review
- Design the first agent with scalability in mind from the start, so it can later extend to outbound calls, internal workflows, ads, and cross-department use
- Don't go through intermediary vendors—connect directly to the source to secure the latest models, lower costs, and direct engineering support
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.