I redesigned JoyzAI's activation around an AI copilot that turns business goals into agents, workflows, and CRM setup.
Role
Platform
Status
Year
Joy is an AI Sales Agent that helps small and medium businesses automate sales across WhatsApp and voice calls using AI, CRM, and workflow automation.
As the platform evolved, it expanded beyond AI agents to include workflows, CRM, and knowledge bases. But with that growth came added complexity—users had to configure multiple modules before they could launch their first AI sales workflow and experience any value.
The business wanted customers to go live faster and adopt more platform capabilities, while users simply wanted to describe their sales goals and let AI handle the setup.
Creating an AI sales agent wasn't enough. Users also had to configure workflows, CRM fields, knowledge bases, and other settings before going live.
Users first had to understand how different modules worked together before they could complete a simple task.
Many common tasks required users to repeatedly configure the same settings, making the setup process feel repetitive and unnecessarily time-consuming.
Users thought in business outcomes like "qualify leads" or "follow up with customers"—not in platform features or modules.
To better understand where users struggled, we reviewed onboarding sessions, support conversations, and common setup requests. While each business had different goals, the language they used followed a consistent pattern—they described what they wanted to achieve, not how they expected the platform to work.
Instead of improving individual setup flows, we wanted to rethink how users interacted with the entire platform. These goals became the foundation for every design decision that followed.
Generate AI agents, workflows, reports, CRM configurations, and other repetitive setup automatically, allowing users to review and refine instead of building everything from scratch.
Let users describe what they want to accomplish in natural language while Joy Copilot identifies the right capabilities and performs the required actions.
Unify creation, search, reporting, product guidance, and support into a single conversational experience, eliminating the need to navigate between different modules.
Rather than improving individual features, we focused on the decisions that would simplify the entire product experience.
01
Users should describe goals—not configure systems
People naturally think in business outcomes, not AI Agents or Workflows. Starting with intent removes the need to understand the platform before getting value.
02
AI should do the heavy lifting
Creating agents, workflows, reports, and configurations manually slowed users down. Letting AI generate the first draft reduced setup effort and accelerated time to value.
03
AI should guide, not replace
Users remained in control by reviewing and refining every AI-generated output, building trust without sacrificing the speed of automation.
04
Support should begin with AI
Most support requests were repetitive setup questions. Resolving them instantly through Joy Copilot reduced support dependency while allowing human agents to focus on critical issues.
01
02
03
04
