This project was focused on fixing a broken onboarding and validation experience in a compliance and regulatory platform. New users struggled to complete onboarding because the process was long, confusing, and heavy on regulatory jargon. They had to jump across screens, upload unclear documents, and often redo steps due to errors or missing information.
To solve this, I redesigned onboarding around an AI-powered conversational agent that guided users step by step, validated inputs in real time, and removed guesswork from the process.
Role : Lead Product Designer
Project Duration : 1 month
Tools Used : Figma
The Challenge
Onboarding process for this compliance and regulatory application was long, time consuming and cumbersome, Users had a tough time completing the process because of the following reasons.
- Long, form-heavy workflows
- Too many documents to upload, with no clarity on format or content
- No save-and-resume option
- Free-text inputs that created messy, inconsistent data
- Jargon-heavy language that confused non-experts
- Back-and-forth communication to fix incorrect or incomplete profiles
Users needed:
- Guidance, not more instructions
- Validation, not assumptions or guesswork
- A way to move forward without feeling stuck or unsure
So instead of redesigning another form-based flow, I reframed onboarding as a conversation.
The Turning Point: Introducing an AI Agent
I designed an AI-powered chat agent to act as a guardrail, not an autopilot. The agent didn’t replace the user’s responsibility. It supported it.
Here’s how the experience changed:
- The agent pre-filled company and profile information where possible
- Users reviewed and edited instead of starting from a blank screen
- Information was presented in small, focused steps
- Progress and remaining steps were always visible
- Users could save and resume anytime
If something went wrong, the agent explained why—in plain language
Instead of wondering, “Am I doing this right?”, users finally felt confident moving forward. Every decision focused on one question:
“What does the user need to feel confident enough to continue?
How I defined the AI agent.
How I defined the AI agent.
- Defined a AI agent persona with a calm, supportive tone
- Structured the conversation into repeatable patterns:
- Opening
- Request
- User response
- Validation
- Confirmation and next step
We built a design system to unify the experience and support future growth. Teamed up with training and change management to tailor rollout support across teams. Finally, we added automated updates throughout the ticket lifecycle so users stayed informed without needing to ask.
The redesigned onboarding replaced long, form-heavy workflows with a guided AI-agent experience, significantly improving efficiency and data quality.
Key improvements:
- Reduced completion effort from 16–18 interactions to fewer than 8 guided steps
- Shortened onboarding time by replacing long forms with focused conversational inputs
- Fewer user errors due to real-time validation and structured responses
- More consistent, standardized data captured during onboarding
- Improved dashboard accuracy from cleaner profile inputs
Overall, users moved through onboarding faster, with fewer mistakes and greater confidence in their submissions.
