AI agents

Tapcheck is an earned wage access platform. Mastercard approvals, a core revenue driver, required payroll managers to review and approve employee requests through the employer portal.

I led product design for Tappy Agent, an AI-powered assistant embedded in the portal. 2-month project with 2 FE, 2 BE engineers and a PM.

Payroll managers found approvals stressful and time-consuming, often delaying or skipping them when information wasn’t clear.

• Approvals scattered across multiple screens

• No way to ask clarifying questions without contacting support

• Incomplete data caused batch abandonment

Centralized approvals into a conversational single-container flow. Tappy Agent surfaces only the key info managers need and lets them ask natural language questions about employees.

Introduced Approve all, Review individually, Do it later, and Ask a question as the core action set. Phased rollout: Phase 1 shipped core actions, Phase 2 adds pre-built Q&A buttons from logged requests.

Explored multiple prototyping tools to validate the conversational flow quickly and find the right fidelity for user testing.

The final design centralized every approval action into a single conversational container with clear data hierarchy and natural language Q&A.

• 14% increase in batch approval completions

• Approval time cut from 10 days to 6 days

• 10% reduction in CS tickets

• Increased Mastercard adoption (core revenue driver)

Users said Tappy Agent felt like a real assistant, not a redirect tool.