How I Unified Drift Management User Flows Across 11 Products with AI-in-the-Loop

User Flows Design

Design System

AI-in-the-Loop

~8 mins Read

Context

Drift is the gradual deviation of an IT environment from its intended state. Many products in my company needed to help users manage drifts, but each approached it differently.

Context

Drift is the gradual deviation of an IT environment from its intended state. Many products in my company needed to help users manage drifts, but each approached it differently.

Context

Drift is the gradual deviation of an IT environment from its intended state. Many products in my company needed to help users manage drifts, but each approached it differently.

Context

Drift is the gradual deviation of an IT environment from its intended state. Many products in my company needed to help users manage drifts, but each approached it differently.

Context

Drift is the gradual deviation of an IT environment from its intended state. Many products in my company needed to help users manage drifts, but each approached it differently.

Why This Matter?

While our design system ensures visual consistency, it doesn't define how users complete tasks. This creates friction in Drift user flows that should feel familiar across products.

Why This Matter?

While our design system ensures visual consistency, it doesn't define how users complete tasks. This creates friction in Drift user flows that should feel familiar across products.

Why This Matter?

While our design system ensures visual consistency, it doesn't define how users complete tasks. This creates friction in Drift user flows that should feel familiar across products.

Why This Matter?

While our design system ensures visual consistency, it doesn't define how users complete tasks. This creates friction in Drift user flows that should feel familiar across products.

Why This Matter?

While our design system ensures visual consistency, it doesn't define how users complete tasks. This creates friction in Drift user flows that should feel familiar across products.

My Contributions

Unified Drift management user flows
Integrated AI to accelerate the design process
Defined key human-AI touch points
Exploring AI-in-the-loop design pattern

My Contributions

Unified Drift management user flows
Integrated AI to accelerate the design process
Defined key human-AI touch points
Exploring AI-in-the-loop design pattern

My Contributions

Unified Drift management user flows
Integrated AI to accelerate the design process
Defined key human-AI touch points
Exploring AI-in-the-loop design pattern

My Contributions

Unified Drift management user flows
Integrated AI to accelerate the design process
Defined key human-AI touch points
Exploring AI-in-the-loop design pattern

My Contributions

Unified Drift management user flows
Integrated AI to accelerate the design process
Defined key human-AI touch points
Exploring AI-in-the-loop design pattern

Define Problem

Finding What Our Design System Missed

Uncovering a Hidden Gap in Drift Management

While designing the PowerFlex upgrade experience, I discovered there was no clear way to manage drifts after upgrades, leaving users to rely on personal experience.

As a System Engineer, I need to know what, why, and how something drifts in my system.

Spotting Inconsistencies Across Products

I reached out to other product teams for reference but found each had its own way of handling drifts. That’s when I realized a key gap in our design system: It kept visuals consistent but ignored how users actually completed tasks.

Insights from interviews with design leads across products. Details are intentionally vague due to NDA.

Why Solving This Wasn’t Simple...

01

Fragmented Flows

01

Fragmented Flows

01

Fragmented Flows

02

Limited Time

02

Limited Time

02

Limited Time

02

Limited Time

03

Unclear Strategy

03

Unclear Strategy

03

Unclear Strategy

03

Unclear Strategy

Section 01

How I Defined a Unified Framework for Drift Management

Challenge

Each product had its own goals and legacy flows for handling drifts.

Challenge

Each product had its own goals and legacy flows for handling drifts.

Challenge

IT teams had to switch between 5+ siloed systems just to track down an issue.

Challenge

Each product had its own goals and legacy flows for handling drifts.

Finding the Shared Pain Points

To bring clarity, I met with design leads from 11 products and reviewed their drift user flows. Despite different contexts, 3 recurring pain points emerged:

Inconsistent Terms
Manual Executions
Relying on Personal Experiences

Defining 3-Phase Framework and 9 Main User Tasks

Section 02

How I Partnered with Design Leads and Used AI to Accelerate Process

Challenge

Tight timelines left little room for research and cross-team alignment.

Static, overly technical visuals left users struggling to see how things were connected.

Teaching AI to Think Like a Designer

My first attempt was simple: feed insights into Gen-AI tools and hope for quick answers. The output was equally simple - fast but generic. AI flattened nuances and missed context.

That’s when I shifted my approach. I treated AI like a junior designer needing direction: rewriting prompts with context and examples and asking AI to simulate user perspectives.

By reviewing each AI-generated user story and testing my assumptions against research, the results evolved from surface-level summaries to deeper insights showing real decision logic and role differences.

Early experiments with FigJam’s embedded AI.

An example of how I applied AI to analyze research insights.

Reviewed with Researchers and Design Leads

I validated these insights with UX researchers and design leads from each product. Their feedback helped refine the findings into actionable design directions, ensuring speed didn’t compromise accuracy or alignment.

Reviewed the draft user flows with researchers and design leads.

Section 03

Extending the Project: Integrating AI into User Flows to Enable Human–AI Collaboration

Challenge

Every product team wanted to integrate AI but lacked a clear guidance

Challenge

Every product team wanted to integrate AI but lacked a clear guidance

Challenge

Every product team wanted to integrate AI but lacked a clear guidance

Challenge

Every product team wanted to integrate AI but lacked a clear guidance

Challenge

Every product team wanted to integrate AI but lacked a clear guidance

Choosing the Right Type of AI: Workflows vs Agents

Not every AI idea is practical. To stay grounded, I compared AI Workflows and Agentic AI.

In high-stakes environments, reliability outweighs novelty. So I chose the AI Workflow approach, focusing on:

Drift Detection & Classification
Impact Prediction & Resolution Advice,
Post-Action Summaries & Documentation.

Designing Human-AI collaboration in the Flow

I designed AI as a collaborator, not an add-on. Each touch point clearly defined what AI assists with and what the human owns:

AI explains, predicts, and drafts.
Humans review, refine, and decide.

"I detect and classify drifts, explain what changed, assess the risk, and suggest who should act."

Monitor

"I simulate potential system impacts and recommend resolutions based on historical data."

Analyst

"I summarize what happened, identify who took action, and update the baseline when change is intentional."