Client
Careem
Challenge
Transform a powerful backend AI system into an intuitive product for customer care teams with varying levels of technical expertise.
Impact
An AI Agent Builder design that empowers customer support teams to automate workflows, cut manual tasks, and resolve issues faster.
Koos excels at translating complex, technical logic into human-centered simplicity. In only six weeks, they gave us the blueprint and clickable prototype we needed to confidently launch and scale our AI agent builder.
An intuitive interface to automate AI customer support
The result of our collaboration is an AI Agent Builder that empowers Careem’s employees to automate and manage their customer support workflows. We developed a clickable prototype that demonstrates the core functionality, providing a tangible vision of the final product. This user-first interface enables customer support employees to visually construct and optimize sophisticated AI agents by connecting predefined building blocks.
This design not only makes powerful AI accessible but also fosters a shared understanding, creating a scalable system that can evolve with business, employee and customer needs.
Scalable UX architecture
A consistent information hierarchy supporting usability and long-term product growth.
White-Label design system
A flexible, reusable UI component library built for rapid implementation.
Service blueprint
A visual map to align user activities with AI system responses for an effective Human-in-the-Loop experience.
Clickable high-fidelity prototype
An interactive, desktop-first prototype demonstrating core user flows and accelerating stakeholder validation and MVP planning.
Making enterprise AI usable for everyone
The real challenge was not the technology itself but usability. Careem’s AI already delivered strong automation results, but to scale impact, the experience had to work for customer-care teams with diverse technical skills. The goal was to design a human-centered interface that simplified complex logic and encouraged cross-functional collaboration, turning sophisticated AI capabilities into everyday tools for support teams.
From service blueprint to clickable prototype in 6 weeks
Koos began with a service blueprint to align technical and business perspectives, mapping internal workflows and identifying friction points. Building on this shared understanding, the team translated backend concepts like webhooks, intents and fallbacks into a drag-and-drop visual builder. Rapid design sprints produced a clickable, high-fidelity prototype that demonstrated core user flows, enabling early feedback and reducing risk before MVP development. All of this was completed within a six-week collaboration.
Next steps
This collaboration laid a strong foundation for scaling AI in customer support. With a scalable design system, UX architecture, and Service Blueprint in place, the next phase will focus on validating the concept and moving toward production. Key next steps include:
- Validating the prototype with real users to gather feedback, improve usability, and confirm product–market fit.
- Designing for edge cases and exception scenarios to ensure the product handles errors, incomplete inputs, and unexpected behaviors, resulting in a robust, enterprise-grade experience.
- Developing a fully functional MVP to bring the AI Agent Builder into live use
- Expanding functionality for enterprise environments to enable more teams to independently build, launch, and manage AI-powered support workflows
