Dialpad Product Integration Case Study

Dialpad

AI conversations for customers and support agents

Integrated Koopid’s AI workflows into Dialpad.

What reached production

Shipped with conversation context and live-agent handoffs preserved.

My role: Product Designer
Timeline: 2021–2022 · 8 months
Product: Dialpad integration · Shipped

Problem

Bringing our product into theirs.

After Dialpad acquired Koopid, I moved over to bring our AI workflows into Dialpad’s platform. The goal was to make them feel native to Dialpad without losing what made them work: the customer’s history and the handoff from automation to a live agent.

The Dialpad desktop client with a customer conversation open — The ecosystem the workflows had to move into.

Research

First, find the overlap.

Before touching any visuals, I inventoried our workflows, permissions, states, and owners, then compared them with what Dialpad already had. Where Dialpad had an equivalent, we used theirs. Where we had something they didn’t, I reshaped ours to follow their design system rules while keeping its purpose intact.

Dialpad Digital Engagement queue beside the active customer conversation — The inventory covered the queue, the thread, and the states in between.

Decision

A handoff.
Not a restart.

I translated the workflows into Dialpad’s patterns, keeping context connected across automation, messaging, and agent tools.

One conversationHistory travels with it

Dialpad customer-facing chatbot interface

Start with what the customer needs.

The reason for escalation follows the customer from automation to the agent.

Decision · The piece I fought for

The connector builder had to stay.

The workflow connector builder was the feature I pushed hardest to keep. It gave teams the flexibility to build more complex IVR flows, and it let Dialpad’s engineers hook directly into our models. Think of it as a small version of what MCP does today.

Decision · Migration

What to keep, adapt, or retire.

The progression from original Koopid UI to the redesigned Koopid experience and native Dialpad interface — One capability, translated across product generations.

Keep.

Protect the workflows and interaction logic customers depended on.

Adapt.

Use Dialpad’s navigation and components; review new states with product and design-system partners.

Retire.

Remove duplicate patterns that added maintenance without helping customers.

Approvals

Twenty to forty designers, many approvers.

The hardest part wasn’t the design, it was approval. Dialpad had 20 to 40 designers working in pods, and who signed off on a component depended on which pod owned it. Mapping owners early paid off. I aligned product, design-system, engineering, and QA partners before migrating anything, and brought each change to the right person with the reasoning written down.

Outcome

It shipped with the conversation intact.

AI workflows reached production with conversation history and agent handoffs preserved. The existing component foundation adapted to Dialpad’s system, and I updated Dialpad’s Storybook library with the new patterns. I also helped run design hackathons with junior designers and interns. We’d pair up, build a component together as a fun exercise, then walk through how it would be used in real work.