Einstein Validator Salesforce UI screenshot
Einstein Validator Salesforce UI screenshot

I defined the interaction model for an AI-assisted order workflow, establishing how users review, correct, and approve AI output before Salesforce takes action. The concept placed third in a company-wide innovation competition.

I defined the interaction model for an AI-assisted order workflow, establishing how users review, correct, and approve AI output before Salesforce takes action. The concept placed third in a company-wide innovation competition.

Einstein Validator Salesforce UI screenshot

My Role

Lead Product Designer and Product Owner

Timeline

Oct 13, 2023 - Jan 17, 2024

Team Members

Team of five: 1 product designer/product owner, 1 other product designer, and 3 technical architects

Recognition

Third Place

Problem Statement

Employees had to manually read customer emails and re-enter order details into Salesforce. This process was repetitive, time-consuming, and vulnerable to data-entry errors.

The Design Question

How might we use AI to reduce manual work without removing human control?

Because incorrect data could affect real customer orders, I framed a human-in-the-loop workflow:
AI extracts. The user verifies. Salesforce executes.

My Role

As Lead Product Designer and Product Owner, I:
Defined the product concept and user journey
Aligned and led a cross-functional team of five
Designed the AI assistant and validation flow
Shaped the prototype and presentation narrative
Presented the final solution to the judging panel

Meet Einstein Validator

Meet Einstein Validator

Einstein Validator analyzes incoming emails and converts the information into structured order details.

Before creating the order, users can review and confirm the AI-generated information.

This reduces manual entry while keeping employees in control of the final decision.

Our Einstein Validator improves all of these areas: accessibility, privacy, product promotion, deal generation, order creation, reviews, and more.

1. The customer submits an order request by email. 📧

Gmail order workflow screen
Gmail order workflow screen
Gmail order workflow screen
Gmail order workflow screen

2. Einstein Validator extracts the order details and allows the seller to review them without manually re-entering the information. 🤖

3. Once verified, the order is created and stored in Salesforce. 📦

Salesforce orders screenshot

VERIFIED OUTCOME

Outcome

Third-place recognition in a company-wide Salesforce innovation competition.

PROJECTED VALUE — NOT A MEASURED RESULT

⏱️

Designed to cut the time and error risk of manual order entry from unstructured emails.

Human-in-the-loop validation before Salesforce executes any order.

🔒

User control preserved — no order is created without explicit confirmation.

PROPOSED NEXT STEPS — NOT COMPLETED WORK

If this concept moved beyond the competition, the human-in-the-loop review step could go further: confidence indicators on each extracted field, highlighting for fields the AI is uncertain about, a dedicated correction flow for fixing extracted values before confirming, and error recovery if Salesforce order creation fails. None of this was built for the competition submission — it’s a direction the human-in-the-loop model naturally opens up.

Testimonial

Janie’s ability to coordinate, encourage, and motivate a busy team of people was a key success factor that enabled the team to complete the solution within the competition timeframes.

She showcased her exceptional ability to visualize the solution from the user’s perspective and describe the value in terms of accuracy and time saved. Janie’s presentations were of top quality. Thanks to her efforts, the team won third prize in the competition.

Mathanan Yogaratnam - VP, Delivery Leader at Salesforce

Mathanan Yogaratnam
- VP, Delivery Leader at Salesforce

Hackathon presentation image

Janie Park, Lead Product Designer based in New York

©2026

Janie Park, Lead Product Designer based in New York