Agentic Onboarding
How I reduced turnaround time by 50% and increased feature adoption among new users.
Conversational Design
AI Agents
Explainability
Shipped
Note: Designed for a leading New York-based wealth and asset management firm. This is a white-labeled adaptation of the original product design.
INTRODUCTION
I designed an end-to-end agentic onboarding experience for institutional KYC investigations, replacing static upload forms and RM-dependent workflows with intelligent agents that parse regulatory documents, validate information, and guide clients through due diligence in natural language. I defined the agent personas, conversational architecture, and orchestration flows to deliver personalized guidance while reducing manual intervention and making complex compliance workflows more efficient and scalable.
What I achieved :)
Decrease in Document Upload Support Tickets
from new users completing the onboarding
18%
Increase in Task Completion Rate
among existing users for KYC refresh
50%
Decrease in Document Intake Time
reduced from minimum ~36 hours to avg ~12 hours
THE DESIGN
Agentic document upload and data validation in a conversational modality
A conversational way to complete the upload process
Bringing guidance and clarification directly into the onboarding experience through an AI agent that communicates in natural language and supports clients at every step of document submission.
A flexible system that adapts to each entity type
Replacing the static, one-size-fits-all experience with a flexible system that adapts to each client’s entity type and generates a personalized document checklist.
Agentic document upload, matching, and data validation
Designed an agentic bulk-upload experience that takes the manual work out of document submission. Clients can upload multiple documents at once while agents automatically scan, classify, and match them against requirements, then validate key data points. This saves time and effort of internal teams, contributing to that reduced turnaround time.
In-chat clarification for mismatched or missing information
Parsed documents are cross-checked against benchmark values and one another to detect inconsistencies. When a mismatch is found, the agent flags it and prompts clients for clarification directly in the chat, resolving issues as they arise rather than through lengthy email exchanges. This reduces back-and-forth and contributes to the 50% reduction in turnaround time.
Human-in-the-loop for edge cases we could not map
When the agent encounters an exception that requires human judgment or falls outside its defined validation rules, the workflow doesn't force an automated decision or leave the client stuck. Instead, the case is flagged and routed to the appropriate internal team member through their portal for review and resolution. This creates a clear handoff between AI automation and human oversight, allowing complex or ambiguous cases to move forward without compromising the accuracy of the KYC process.
CONTEXT
A blocker in scaling multi-agent KYC and due diligence workflows
The client wanted to reduce KYC turnaround time by using agentic workflows to automate manual tasks across 20+ internal due diligence teams. But there was a critical dependency: those workflows could only begin once clients submitted their documents through the customer portal. This made portal usage essential. It was no longer just a place to upload documents; it was the starting point for the entire agentic ecosystem.
THE PROBLEM
The portal already existed. Nobody used it.
Clients emailed documents to their Relationship Manager instead.
For clients this means
Got asked for the same document more than once
No guidance on what was needed or why, had to rely on RM availability
For internal teams this means
No central document set, so due diligence couldn't start until someone chased it down
Manual coordination across 20+ teams to figure out what had been collected
A blocker on agent-ification of due diligence process due to low adoption rates
RESEARCH
It wasn't the modality clients rejected, It was the rigidity.
Research showed that clients weren't opposed to using digital tools. They relied on Relationship Managers because they offered something the portal couldn't: context, guidance, flexibility, and the ability to handle exceptions.
The existing portal treated every client the same. The challenge was to bring that same level of guidance and adaptability into the digital experience, rather than simply making document uploads easier.
I usually just send everything to my RM. They tell me what I actually need and what I can use instead."
OUTCOME
An onboarding that people could trust
57%
Decrease in Document Upload Support Tickets
from new users completing the onboarding
18%
Increase in Task Completion Rate
among existing users for KYC refresh
50%
Decrease in Document Intake Time
reduced from minimum ~36 hours to avg ~12 hours
Other Project Outcomes
Build trust in AI
Customers understand and confidently act on Sophia’s recommendations.
Increase first-pass completion
More submissions completed without corrections or follow-up.
Reduce customer effort
Fewer manual inputs, uploads, and repeated requests.
What I Learned
This project reinforced that adoption is a design problem before it's a feature problem. The portal already existed, and the fix wasn't adding more to it but designing in the flexibility clients had with their RM.
It also sharpened how I think about AI in regulated contexts: deciding what Clara would never do was as deliberate a design decision as what it would, and getting that boundary right was what made the system trustworthy enough to ship.
Most importantly, it was a reminder that designing for one user group in a complex system is rarely enough. The client-facing flow only worked because it also solved the internal teams' problem, and adoption without that would have been a vanity metric.
