Grab launched a 3-month company-wide AI sprint. Everyone had to pause BAU and work on an AI initiative.

Design submitted dozens of problems in a shared Figma Jam board. 24 contributors across Singapore, Indonesia, Malaysia, and India picked the Research Automation topic.

There was:

• No workflow

• No alignment

• No squad structure

• No decision-making model

• Conflicting expectations from leaders

• Limited engineering availability

I was chosen to lead this entire research topic team.

A large team running in 24 different directions with no real output

A complex, mixed group of 24 contributors with:

• Different markets, cultures, seniorities

• Dozens of overlapping problem statements

• Unclear goals

• BAU conflicts

• High expectations but no structure

• Mid-sprint leadership direction changes

Turned chaos into a clear problem space

I synthesised the full set of problems into an end-to-end research workflow map.

Then I ran severity × frequency discussions and async voting across all 24 members to agree on four problem themes

This moved the team from fragmented noise to a shared understanding of where the biggest pains were.

Built the operating model from zero

Because this sprint team had no structure, I created the entire operating system:

Rituals

• Each squad set their own recurring syncs

• 1×/week full-team sync

• Async update and discussion through Slack & FigJam

Decision model

• Diverge → vote → converge

• Clear scoping templates

Alignment loops

• Weekly demos to design, product, and engineering leaders

This gave our cross-country team a predictable rhythm.

Formed and aligned the squads

The themes became four squads:

1. Research Ops Automation

2. Insight Synthesis Automation

3. Insight Accessibility

4. UT Automation

I positioned contributors based on interest and capability.

Each squad received:

• A clear problem statement

• Boundaries

• Rituals

• Template for demonstrating value

Navigated a major leadership shift mid-sprint

Midway, leadership recommended killing Insight Accessibility to shrink the team from four squads to three.

After negotiation, they told me:

“Bring it down from four squads to three —
you decide which squad gets cut.”

I didn’t cut Insight Accessibility.

Instead, I made a structural move:

• Merged Research Ops Automation + Insight Synthesis Automation into one larger, coherent squad

• Preserved Insight Accessibility

• Reframed the overall direction into an integrated long-term system

• Protected morale, motivation, and ownership

• Aligned leaders without conflict

Outcome

The squad leadership initially wanted removed — Insight Accessibility — was later chosen to present at EXCO during the sprint.

Ensured delivery, visibility, and continuation

By the end of the sprint:

• 3 squads delivered usable POCs and prototypes

• 1 squad presented at a company brownbag session

• Work was featured in the company newsletter

• 2 squads gained visibility through EXCO presentations

• The research automation work continued after the sprint— sustained by motivation, not leadership push

This continuation happened because the team cared deeply and believed in the value, not because they’re mandated to.

Structural Impact

• Built the complete operating model for our 24-person, multi-country team

• Turned dozens of messy ideas into a coherent problem space

• Created rituals, alignment loops, and decision models for predictable execution

• Stabilised progress despite scope shifts and external pressure

Leadership Impact

• Balanced four countries with different working styles and opinions

• Protected a squad scheduled for removal without breaking leadership trust

• Made hard structural decisions to maintain momentum and morale

• Kept contributors motivated even after leadership support faded

• Created psychological safety and focus in a chaotic environment

Organisational Impact

• Research Automation and Insight Accessibility squad presented to EXCO

• Featured in design org newsletters

• Presented in design brownbag sessions

• Other company requested demos

• The work became the basis for the long-term research automation initiative

Delivered inside a large, diverse, ambiguous AI sprint team

• Created alignment across markets

• Brought order to ambiguous initiatives

• Structured teams for impact

• Protected momentum through leadership shifts

• Managed cross-functional complexity

• Balanced human and strategic decision making