Design Director
Selected work / 01
Back to portfolio ↗
Twilio · Conversation Intelligence
Conversation
Intelligence
Turning conversations into intelligence
I led design across Voice & Video, Flex and Conversations to help turn conversation data into something teams could understand, configure and act on.
The opportunity
From conversation data to useful intelligence.
Conversation Intelligence sits at a critical point in the Twilio experience. It takes what is happening in customer conversations and turns it into structured insight that can support better decisions and actions.
The challenge was not simply designing another feature. It was making a sophisticated AI capability feel understandable inside a platform used to build and operate real customer experiences.
The real design challenge was making the intelligence useful without making the complexity visible.
Leadership scope
One product problem. Multiple teams.
As Senior Product Design Manager, I brought together design perspectives across Voice & Video, Flex and Conversations to solve a shared problem around conversational intelligence.
This meant aligning designers, Product and Engineering around a common experience while still respecting the different workflows and technical realities of each product area.
What made this a design leadership problem
01 · Align
Bring Voice & Video, Flex and Conversations into one design conversation.
02 · Model
Define how intelligence should work across the product.
03 · Simplify
Turn complex AI configuration into understandable interaction.
04 · Scale
Create patterns and quality practices the teams could reuse.
My role
A portfolio problem, not a single-product problem.
Conversation Intelligence brought together work that crossed the teams I led across Twilio Communications and Conversational AI.
That meant connecting perspectives from Voice & Video, Flex and Conversations, while partnering closely with Product and Engineering across the portfolio.
Rather than treating each surface as an isolated experience, I used Conversation Intelligence as an opportunity to create a more coherent model for how intelligence could move through the wider Twilio platform.
The product
Conversation Intelligence turns conversations into signals people can use.
The product needed to do more than surface AI results. Customers had to be able to configure how intelligence was generated, preserve the logic they already had, and understand how those settings affected real conversations.
That meant designing across three connected layers: the intelligence itself, the configuration that drives it, and the conversation experience where the results become useful.
02 · Configure intelligence
Make the intelligence itself understandable and editable, including customer-authored operators and richer training examples.
03 · Turn insight into action
Connect intelligence to rules, triggers and actions so the output can drive what happens next.
The design challenge
How do you make AI intelligence understandable without exposing all the complexity underneath?
CINTEL was evolving quickly, including a v2 to v3 migration, new configuration capabilities, editable code, training examples and additional rule triggers. The experience had to evolve without making customers relearn everything.
The experience
See the product, then explore the work behind it.
The walkthrough shows the CINTEL experience in context. The Figma work below shows how the team explored the product model, interaction patterns and migration experience.
Conversation Intelligence walkthrough
AI · Product design · Configuration
CINTEL design work
01 · Figma
Conversation Intelligence design file ↗
The main CINTEL design work, including the product experience and exploration.
02 · Keyframe
Keyframe exploration 01 ↗
A wider journey exploration showing how the experience was worked through across states and interactions.
03 · Keyframe
Keyframe exploration 02 ↗
A second exploration of the CINTEL experience and its relationship to the wider conversational workflow.
04 · Prototype
Conversation Intelligence prototype ↗
The interactive prototype starts at the CINTEL migration experience.
CINTEL in context
A product shaped across Voice, Flex and Conversations.
CINTEL was not designed in isolation. I brought together the teams and product perspectives I was leading across Voice & Video, Flex and Conversations to create a coherent experience for conversational intelligence.
The architecture below helped make the relationships between intelligence, conversations, memory, knowledge and communications explicit, while keeping CINTEL itself at the centre of the user problem.
Sierra v1 UX architecture
The product context context around Conversation Intelligence, used to understand how the product connected to the wider conversational experience.
Design exploration
From product strategy into interaction.
I worked through Conversation Intelligence at multiple levels: the product model, configuration experience, migration journey and the interaction patterns needed to make conversational intelligence understandable in the moment it is used.
01 · Product experience
The CINTEL v3 experience: making configuration, intelligence and action legible inside the platform.
02 · Product model
Exploring the experience and the relationships between the core CINTEL workflows.
03 · Journey exploration
Working through states, transitions and the broader customer journey.
04 · Interaction exploration
Extending the exploration across the wider conversational workflow.
06 · In the conversation
Bring intelligence into the conversation viewer so results are encountered in context, rather than in a disconnected reporting surface.
07 · Prototype
The prototype was where the model was tested as a connected journey, not a collection of isolated screens.
From intelligence to action
The product had to help customers understand what the intelligence meant.
One of the wider design questions was how customers could experience the value of conversational AI before they had invested time configuring it. The Playground work explored a self-serve path from curiosity to an interactive, Sierra-powered experience.
For CINTEL, this reinforced a key principle: the value of intelligence is realised when customers can see how it changes what they can do.
Conversational AI Playground
Supporting exploration · Sierra
Migration
Evolve the product without asking existing customers to start again.
A major part of the work was the v2 to v3 migration. Existing intelligence services had to move into the new model while preserving customer-authored logic and making new capabilities easier to adopt.
The experience evolved around auto-created configurations, migrated operators and a clearer relationship between transcription, conversation and intelligence. Support for multiple training examples became part of the production experience rather than a theoretical future state.
04 · Migration experience
The migration flow preserves the existing customer model while moving users into the v3 configuration experience.
05 · System thinking
The migration work mapped how transcript, recording, conversation and intelligence configurations relate to each other.
Supporting foundation
Transcription became part of the story, not a hidden prerequisite.
The Voice and Recordings work made the capture layer more explicit. That mattered because CINTEL depends on the data being captured, transcribed and routed correctly before intelligence can be generated.
Voice · Recordings · TranscriptionsSupporting the intelligence layer
What I was really designing
A system that could explain itself.
The strongest design move was not adding more UI. It was making the relationships visible: what is being captured, what is being transcribed, what is being analysed, what configuration drives that analysis and where the resulting intelligence appears.
That gave the teams a shared language for discussing the product and gave customers a clearer path through an increasingly complex platform.
Design quality
Quality had to be designed into the way the team worked.
01
Test the idea
02
Test the experience
03
Test the system
04
Test AI behaviour
05
Learn and improve
We established recurring design-quality reviews, pattern QA, UAT and cross-functional testing so the quality of the experience could keep pace with the speed of the product.
Impact
A growing product with a wider role in the platform.
Conversation Intelligence grew into a significant part of Twilio's Conversational AI offering, with enterprise adoption and measurable commercial growth.
$3.63M
Annual revenue associated with Conversation Intelligence.
64%
Year-over-year growth.
3
Design organisations brought into the conversation: Voice & Video, Flex and Conversations.
Leadership
The value of the work was not only the product we shipped. It was creating alignment across teams that had historically owned different parts of the customer experience.
As Senior Product Design Manager, I led 12 designers across the portfolio and partnered with Product and Engineering leadership to align priorities, capacity and design direction.
Conversation Intelligence gave those teams a shared problem to solve: how conversational intelligence should work as part of a connected platform rather than as a collection of separate capabilities.
Further reading
Explore the wider product story.
Twilio · Product launch
What is Conversation Intelligence? Turn live interactions into context-aware action ↗
Twilio · Product launch
Introducing Conversation Orchestrator ↗
Twilio · Product launch
What Is Twilio Conversation Memory? ↗
Twilio · Research report
Inside the Conversational AI Revolution ↗