All Work
IBM × Mondelēz partnership campaign — workshop environment, Bangalore site visit, and Field Guide companion app, all sharing one motion language.

Mondelēz

A printed workbook, rebuilt into a field tool that runs.

Scroll

When the workshop has to keep moving.

Three personas, four lenses each, one workshop day. An opportunity matrix to argue with, an empathy map per practitioner, a session summary that captures everything before the room moves. In 2017 that load traveled on a printed 23-page workbook. Nine years later, I scaled that workbook into a 23-screen field tool, using Claude as a parsing and scaffolding partner, so the same workshop choreography could now run room to room — without a refresh, without a stall.

The 2017 work landed on two surfaces — a partnership microsite and a workshop environment, in Chicago and Bangalore. The Field Guide app is the third surface, retroactively built in 2026 from the original workbook. Same motion grammar across all three; same sense that the technology shows up only as long as it serves.

I led the work end to end — interaction model, motion choreography, visual language, and editorial voice across all three surfaces. Every transition was tuned by feel, then written into a stack a product engineer could fork.

2017 engagement ⬩ 2026 retroactive build ⬩ Field Guide app generated from the original workshop PDF with Claude

Role
Art Direction Motion Direction Interaction Design Creative Direction Producer
Practice
Prototyping AI-assisted build Motion language Interaction storytelling Design-system thinking Code-as-spec Cross-functional partnership Editorial voice
Surfaces
Mobile field tool · 23 screens AI suggestion ritual Matrix placement Per-screen links Partnership microsite Bangalore site-visit environment Embedded device preview
Seven Products, Seven Stories
System

Seven products.
One language of joy.

The insight was structural. Each of the seven service pillars maps directly to a Mondelēz product line — color-coded, named, and tuned so a complex partnership commitment can be read at a glance. The framework above is the spine the rest of the work hangs on: same palette in the microsite, same color-codes in the field tool, same vocabulary in the room.

A framework you can taste.
Microsite

A home that keeps current.

The partnership has had a home since 2017. The microsite frames the brand identity, the seven-product framework, the strategic commitments, and the full story of the Bangalore site visit. Maintained and updated in the years since, the site keeps reflecting where the relationship stands. The same care for typography, motion, and craft shows up in the field tool.

Explore the live site* adapted for public release
A document opens once.
An environment stays open.
Workshop environment in Chicago — placemats, wall graphics, printed decks, and screens all speaking the same visual language.
Bangalore

A handcrafted welcome.

A rangoli spelled out Mondelēz in flower petals at the lobby. A traditional Indian art form, hand-laid, impermanent by nature. The peacock illustration anchoring the environmental graphics was rendered from the teardrop forms of the Mondelēz logo itself, binding the brand mark to the cultural symbol in a single shape. Inside, every engineer carried a trading card with their role, their experience, and their favorite Mondelēz product. Tarun: Oreo. Mainak: Cadbury 5 Star.

Eleven engineers on the welcome board.
Hundreds more in a card in their pocket.
Field Tool

Built for the room,
not the deck.

Three people. Three realities. One framework. The Field Guide app turns the original IBM Design Thinking session into twenty-three screens — empathy mapping, journey, opportunity matrix, session summary, all built to be used on your feet, room to room. Each screen has its own URL so a teammate can jump to Smita’s matrix or Nirmal’s journey in one click. The 2017 site visit ran on print; the app is that 23-page workbook scaled into a working tool.

Run the prototype* self-initiated concept
View the full screen map
Twenty-three screens.
One motion language.
Motion Language

One language,
threaded through.

Three qualities, applied across every surface. Things settle quietly when they change in place. Taps and arrivals land with a small spring. Slower moments glide. The system feels intentional throughout — built once, used everywhere. The matrix tab is where the language lands hardest. Drop an opportunity into a quadrant and it settles with a small spring, the zone answering back. And when the room asks Claude for help, the wait never reads as a wait — a sweep crosses the tray to absorb it, and the suggestions arrive one at a time, each landing like a decision made, not a result returned.

Settle, spring, glide.
Motion lab · three system breaths

Three curves, one system.

Tap any moment to feel its curve. The matrix snap lands with a small spring; the AI wait sweep ticks at a steady pace; the suggestions arrive with a slow glide. Below, all three stacked on the same axis.

Snap into quadrant

Tap an opportunity, then tap a quadrant — the dot springs in with a small overshoot, then settles. The receiving zone glows once to ratify the placement. The spring reads as physical material; the action stamps as deliberate, not incidental.

Duration · 420ms Curve · tap-spring
Open Rajiv’s matrix

AI wait sweep

When the room asks Claude for help, the call takes a couple of seconds. A diagonal light-bar travels across the tray to absorb the wait — work happening, not a stalled UI. The curve underneath glides; the sweep itself ticks at a steady pace, so the rhythm reads as constant labor.

Duration · 1.4s Curve · gentle-out
AI features offline in the public copy

Fly to quadrant

After Claude returns, each suggestion flies into its quadrant in turn — staggered 90ms apart so the verdict reads as a series of decisions, not a flash of new content. The earlier dots simply land; the final one springs in and its quadrant pulses once. Three rhythms in one beat.

Duration · 620ms per chip Curve · arrival-glide
Open Nirmal’s matrix
Same axis · three rhythms
Tap-spring Gentle-out Arrival-glide
Craft

Tuned by feel, built to ship.

The prototype is the spec. Plain HTML, CSS, and JavaScript — no framework, no build step. The motion choices live in the same place as the comments that explain them, so a developer reading the page can take in the whole system at a glance. Every one of the twenty-three screens has its own URL — useful in a workshop, where a teammate can jump straight to Rajiv’s empathy map or Nirmal’s opportunity grid in one click. The handoff is the repo.

Every screen, one click away.
Field Guide app overview — nine narrative sections of the IBM × Mondelēz workshop scaled into mobile screens, from persona introductions through empathy mapping, journey, opportunity matrix, brand provocations, and session summary
System flex · across personas

Three people. Three realities.

Same six-tab system, three distinct work realities — Rajiv on Change Governance, Smita on Skill Management, Nirmal on Demand Forecasting.
The architecture flexes per practitioner, the motion grammar holds.

Empathy map · per persona
Empathy map · Rajiv Mathur · Change Governance Empathy map · Smita Kumari · Skill Management Empathy map · Nirmal Jain · Demand Forecasting
Journey · per persona
Journey · Rajiv Mathur · Change Governance Journey · Smita Kumari · Skill Management Journey · Nirmal Jain · Demand Forecasting
Opportunity matrix · per persona
Matrix · Rajiv Mathur · Change Governance Matrix · Smita Kumari · Skill Management Matrix · Nirmal Jain · Demand Forecasting
Workflow

AI in the workflow.

I made every choice; AI sharpened many of them. The starting point was a 23-page printed workbook from the 2017 engagement — three personas, thirteen journey stages, forty-two empathy quotes, the seven-product framework, four AI-era provocations. Claude read the PDF into a structured outline, proposed how the screens should connect, and laid in a first pass. I shaped the architecture, tuned the motion, and wrote every line of the build. 23 pages in, 23 screens out, 320+ data points carried. The AI suggestion sequence went through three iterations before the timing felt right; Claude noticed the suggestion was landing too abruptly, and I rebuilt it so the verdict reads as a decision arriving, beat by beat. AI as the design partner you brief on intent and review for craft — accelerant, not author.

A workbook, made interactive.
Next Project

View all work

Explore