PepsiCo

Designing an internal AI system to reduce cognitive load and improve decision-making for global security teams.

As a UX Design Intern for PepsiCo, I worked on an internal information security initiative focused on reducing cognitive load and improving decision-making for analysts operating in high-stakes environments.

Timeline

May 2025 — August 2025

Role

UX Design Intern

Team

UX Manager, Information Security Lead, ML/AI Engineer, Data Analyst, Project Manager

Tools

Figma, Azure DevOps

Context

PepsiCo’s Digital Products & Applications team supports global security operations across enterprise systems.

PepsiCo’s cybersecurity organization spans regions and specialized teams responsible for protecting enterprise systems and data. During a 12-week engagement, I was responsible for identifying an opportunity within this ecosystem and designing a prototype solution from the ground up.

Problem

Security analysts investigated threats across fragmented tools, increasing cognitive load and undermining decision confidence.

TISC - ServiceNow

ServiceNow

Flashpoint

ADO

Opportunity

How might we centralize threat data and use AI to support faster, more confident decisions?

Unify cross-platform signals

Combine all sources into one dashboard

Improve triage efficiency

Reduce time to act on threats

Support leadership insight

Increase visibility for leadership and governance

Process

Mapping the Threat Intelligence Lifecycle

I facilitated a cross-functional working session to align on user workflows, risk scoring logic, and pain points. I also used this as an opportunity to hear from cyber security team members directly.

User Research

After conducting qualitative workflow audits and +25 interviews with analysts, I surfaced three patterns

Fragmented Workflows

Analysts switch between four or more tools per investigation, breaking focus and slowing decision-making.

Low-Signal Alerts

Alerts surface without clear urgency or context, forcing analysts to manually assess what actually matters.

No Shared Prioritization

Without a consistent scoring model, severity and urgency are interpreted differently across teams.

Fragmented Workflows

Analysts switch between four or more tools per investigation, breaking focus and slowing decision-making.

Low-Signal Alerts

Alerts surface without clear urgency or context, forcing analysts to manually assess what actually matters.

No Shared Prioritization

Without a consistent scoring model, severity and urgency are interpreted differently across teams.

ExpLoration

Through cross-functional reviews and early feedback sessions, the hybrid approach was selected for further development.

The dashboard reduced tool switching but did not reduce mental effort

The chat improved clarity but lacked macro-level visibility

The hybrid model balanced context and reasoning

Aggregated dashboard

AI chat assistant

Hybrid model

User Journey

The biggest challenge wasn’t lack of data, but the effort required to turn scattered signals into confident decisions.

I mapped the analyst journey to uncover where information security workflows break down, from early monitoring to investigation and response.

User PErsona

User Research

User testing helped us distill the experience to what matters most: one source of truth, faster action, and shared visibility

The solution

SYNTHIA is PepsiCo’s first secure, internal intelligence layer designed to safely correlate data across all of their threat platforms

AI Reseasoning

Intelligent conversational analysis that translates complex signals into clear, actionable insight

Threat Dashboard

A unified, correlation-driven dashboard that surfaces, ranks, and contextualizes threats

Business Impact

Delivered measurable gains in information security effectiveness by reducing noise, accelerating investigations, and making risk visible across teams and leadership.

42%

Reduction in tool switching per investigation

30%

Faster time-to-insight for new alerts

2x

Faster leadership visibility
into high-risk events