Cutting helpdesk tickets by 40% with AI Agents

We designed an AI agent that cut common customer inquiries by 40%, cut response time by 90%, and freed up human agents to focus on high-touch hospitality.

Role Design Director
Client Lettuce Entertain You
Duration 12 weeks
Process Agent Blueprint, App Design
Lettuce Entertain You - Missing Points Self Service

Problem: Humans doing machine work

Diners weren't getting their loyalty points sometimes. So they opened support tickets. Human customer support agents had to manually verify these requests across multiple tools. This process took a long time and tickets piled up. Human agents were unable to detect critical hospitality issues because they were overwhelmed with a growing ticket backlog.

It's a grind.
— Human Support Agent
Missing Points Request - Old Flow

Diners submitted missing points requests through the LettuceEats mobile app (I previously redesigned this app as well).

How can we replace augment humans?

We mapped out the human support agent workflows and identified the repetitive and tedious tasks that an AI agent could automate (and is naturally good at).

Future State Blueprint - showing AI

Surprise: we needed very little AI, most of the automation could be handled via APIs.

AI that delivers value at every step

We proposed building the AI agent in phases that gradually took on more responsibility.

Phase 1: agent validates missing points tickets.

Phase 2: agent validates the transaction.

Phase 3: agent adds the missing points to the customer's wallet.

3-Phase AI Agent Build

Service Blueprinting the 3 AI Agent phases. We also ran a tech spike concurrently to validate integration with Lettuce's existing systems.

Bulletproof design

Right before the build, the client shifted direction: let's allow customers to self-service their missing points inquiries without AI.

Since we had already planned the build in phases and understood the integrations, we simply reordered the phases to build the non-AI parts first.

Self-Service Screen

Users could fill out a form for their missing points. And if the data matched, they'd get their points automatically. If not, they get routed to the AI agent path.

Hospitality, not help tickets

The AI agent took common missing points tickets off the human support agents' plate, allowing them to deliver the signature hospitality that Lettuce Entertain You is known for.

Person using LettuceEats Mobile app in restaurant

Diners were more confident about getting their loyalty points.

PODS Case Study Cover

I led the redesign of the PODS digital experience that drove a double-digit percentage increase in bookings