AgentCloud · Visual Agentic AI Platform

AgentCloud

A visual agentic AI platform for building automated workflows and AI agents using connected data sources.

Role
Product Designer
Duration
2024-2025
Platform
Web SaaS
  • Designed a visual agentic AI platform on top of CrewAI
  • Made data → vector → agent → workflow creation accessible to non-technical users
  • Introduced drag-and-drop workflows and chat-based applications for clarity
  • Rebuilt the product UX to align user intent with backend agent logic

AgentCloud helps users:

  • Connect any data source and convert it into a vector database
  • Build AI agents on top of that data
  • Create chat applications and automated workflows
  • Combine agents, tasks, and processes visually instead of through complex configurations
Product Structure UX Architecture Workflow Design Drag-and-Drop Builder Chat-based UX System Design

Powerful backend capabilities existed, but the UI didn't reflect how users think. Agent creation, data connections, and workflows felt confusing and disconnected.

Issue
Impact
Users struggled to understand how data flows through the system
Confusion about the core product functionality
Agents, tasks, and workflows felt disconnected
Users couldn't see relationships between components
UI didn't reflect how users think about their goals
A strong system that was hard to use
Action
Outcome
Re-structured the entire UX to match user goals, not system internals
Interface aligned with how users think
Designed a clear flow: Data Source → Vector DB → Agent → Workflow → Output
Transparent system architecture visible to users
Introduced a drag-and-drop builder to visualize agentic processes
Complex workflows became easy to understand and build
Designed chat-based applications that operate directly on connected data
Immediate value from connected data sources
Ensured frontend interactions clearly communicated backend behavior
Users understood what was happening behind the scenes

Agent-based systems were still new in 2024, with no established UX patterns

Limited references for how to visually represent agents and vector databases

Needed to support both technical and non-technical users

Required deep understanding of CrewAI concepts to translate them into usable UI

This required R&D, iteration, and close collaboration with developers.

Users found it significantly easier to:

  • Connect data sources
  • Create vector-backed chat applications
  • Build agent-driven workflows

Non-technical users could understand and use the system confidently. Adoption of workflow and agent creation features increased.

The redesigned UX helped align product capability with user expectations. AgentCloud continues to evolve, but the redesigned system established a strong, scalable foundation.

This project demonstrates my ability to:

  • Translate emerging AI systems into usable products
  • Design complex workflows without exposing complexity
  • Bridge the gap between backend logic and human understanding