AI Payment Workflow.
Duration
May 2025 - Present
Location
Palo Alto, CA
Role
UX Engineer
Task
B2B2C Payments
Personalized Offers
Creditor Controls
Connecting consumer behavior, personalized offers and creditor controls into one AI payment workflow
A consumer can only accept an offer the creditor allows, at the moment they're ready to pay. That chain was broken: creditor rules lived outside the product, offers were set by hand, and the consumer portal showed little more than a balance and a pay button. I designed the workflow end to end, from creditor limits to the consumer's checkout.
Key Contributions:
– Drove 4× growth in self-service payment adoption.
– Raised payment conversion from 43% to 78%.
– Designed creditor controls, account-level offers and the consumer portal as one connected system.
IMPACT & VALUE
43% → 78% payment conversion, 4× self-service adoption
By connecting what we know about the consumer, what the creditor allows, and what the consumer sees, payments moved from agent-driven phone calls to self-service.
Highlight:
– Payment conversion 43% → 78%
– 4× self-service payment adoption
– Offers personalized per account without ever breaking a creditor rule
MAIN CHALLENGE
Every discount is a business rule with legal and financial weight, and consumers who owe money are anxious, skeptical, and quick to leave.
A wrong discount is a compliance and revenue problem, not a UI bug. At the same time, every extra step or unclear request on the consumer side costs a payment.
GOAL
Give creditors control, give agents context, and give consumers a path they trust.
My role was to make the rules visible where decisions are made, and invisible where they would only add friction.
Design Progress.
FEATURE 1
Account summary that recommends the next move
An AI summary reads the balance, scheduled payments and last contact, then suggests the next action, for example "consider a follow-up call or a fresh settlement offer," above a recent-activity timeline across SMS, email, calls and notes.
FEATURE 2
Personalized offers inside creditor limits
Settle-in-full, settlement-in-installments and custom plans per account. Sliders for discount and number of payments show the creditor limit inline (for example "Creditor Limit: 40%" or "up to 12 months · minimum $75 per payment"), so agents can personalize without breaking a rule.
Activation dates and channels (Messages, Portal, Voice) control when and where the consumer sees each offer, and the same editor powers both creditor-level and account-level offers so they never drift apart.
FEATURE 3
A payment portal consumers can trust
Low-friction sign-in (last name plus one identifier, with SSN entered in four boxes), a Virtual Assistant that explains why information is needed and states what it will never ask for, and a pay page with clear choices: pay in full, partial payment, or a plan.
FEATURE 4
Closing the loop from outreach to checkout
Offers created by automated letter and SMS journeys appear on the account and in the portal, so what the consumer was promised is exactly what they can accept.
Evaluation.
CLIENT FEEDBACK
Shaped by rounds of client feedback
The portal and offer tools went through multiple rounds of client feedback. I turned each round into a prioritized list, grouped it into shippable changes, and tracked open questions back to the client.
ITERATION
From autosave to deliberate publishing
Offer edits originally saved on every keystroke, which meant a half-finished change could reach consumers. I moved offers to an explicit Save step with a clear Activated state, so agents can work on an offer without it going live.
ONGOING IMPACT
The story does not end here...
Offers now flow from automated outreach straight into the consumer's checkout.
Journeys now create portal offers automatically on their first send, connecting outreach, offer and checkout into one flow.
Reflection.
REFLECTION 1
1. Personalization scales only with visible guardrails
Agents personalized offers confidently once they could see the creditor limit next to every slider. Guardrails in the UI did more than any policy document.
REFLECTION 2
2. Trust is the conversion lever
Explaining why we ask for information, and what we will never ask for, did more for completion than any visual change to the pay button.