All projects

IBM IGX: Redesigning Enterprise Product Lifecycle Planning

Simplified a complex, multi-flow generative AI platform into a step-by-step experience that required zero user training. Turned an internal innovation tool into something teams actually wanted to use.

Year
2024
Client
IBM Consulting
Category
Strategy, UX/UI, AI
Duration
16 weeks
IGX Product Lifecycle Planning dashboard shown on a MacBook Pro in a workspace setting

A powerful platform no one could figure out

IBM Garage had built IGX (Innovation Generation Experience), a generative AI platform designed to help product teams run structured innovation workflows. Data scraping, entity extraction, pain point clustering, idea generation, market fit analysis, and product roadmapping, all powered by AI.

The problem: the platform was built by engineers for engineers. Users faced a wall of configuration options, disconnected workflow steps, and no clear path from "I have data" to "I have a validated product idea." Adoption was low. Teams that tried it needed hands-on training sessions just to get started.

The ask was clear: redesign the experience so that any product owner, strategist, or R&D team member could walk in, select their role, and launch a guided AI workflow without prior training.

Pain points
  • Wall of configuration options with no clear starting point
  • Disconnected workflow steps across 7 AI workflow types
  • No guided path from raw data to validated product idea
  • Every new team required hands-on training to get started
  • 7 roles, 4 phases, all shown at once regardless of context

Product strategy through detailed UI

I owned the end-to-end product design: information architecture, interaction design, and visual design using IBM's Carbon Design System. I worked directly with the engineering team and product stakeholders to define the UX strategy, then designed every screen from the landing dashboard through individual workflow steps.

This wasn't just a UI reskin. The fundamental information architecture needed to change. The existing platform presented all capabilities at once. The redesign introduced role-based filtering, phase-based navigation, and progressive disclosure so users only saw what was relevant to their work at each step.

Responsibilities
  • Information architecture
  • Interaction design
  • Visual design (Carbon Design System)
  • UX strategy definition
  • End-to-end screen design
  • Engineering collaboration

From wall of options to guided flow

Step 01
Restructure around roles and phases
The existing platform showed every workflow to every user. I designed a filtering system based on role (Product Owner, Strategist, Solution Architect, Designer, Quality Engineer, AI SME) and phase (Exploration, Conceptualization, Realization, Evolution). This immediately reduced cognitive load: a Product Owner in Exploration sees only the three workflows relevant to their work.
Step 02
Separate workflows from outcomes
The original UI mixed process workflows (how you get there) with outcome tools (what you produce). I split these into distinct sections: Recommended Workflows and Recommended Outcomes. Workflows like Value Orchestration, Golden Thread, and Product R&D sit alongside outcome tools like Pain Point Tracker, Value Based Backlog, Feature Roadmap, and Value Dashboard.
Step 03
Design the step-by-step flow
Each workflow follows a consistent pattern: add data sources, process with AI, review generated outputs, refine through chat, and export. The left sidebar acts as a progress indicator, showing completed steps (Data Scraping, Entity Extraction, Pain Point Clusters) and upcoming ones (Market Fit, Overview). Users always know where they are and what comes next.
Step 04
Make AI collaboration visible
AI-powered steps are tagged with a clear "AI" badge throughout the interface. The Idea Refinement step uses a conversational Watson assistant that guides users through refining generated ideas, showing reasoning and surfacing references. This makes AI feel like a collaborator, not a black box.
Step 05
Enable workflow customization
Power users needed the ability to build custom workflows. I designed a kanban-style workflow builder where users can drag workflow steps across the four product phases (Exploration, Conceptualization, Realization, Evolution) and compose their own pipeline. Preconfigured templates provide the starting point; customization provides the depth.

Key screens

IGX Product Lifecycle Planning dashboard with role and phase filtering
Product Lifecycle Planning dashboard with role-based filtering and AI-enabled workflow toggle

Organizing dense information within Carbon constraints

Every screen in IGX carried a heavy information load: workflow modals needed to surface overviews, target roles, AI capabilities, expected outcomes, and example output templates in a single view. Idea detail pages went further with expanded summaries, color-coded pain point tags, competitive landscapes, recommended actions, potential partners, reference documents, and full refinement histories. All of it had to fit within Carbon Design System components without feeling overwhelming.

The solution was a consistent two-column layout with color-coordinated tags and clear typographic hierarchy. The left column walks users through context and capabilities. The right column surfaces structured data: outcomes, competitors, partners, and links. Users can scan without reading everything, and the color system makes related information instantly groupable.

Product R&D workflow detail modal showing overview, outcomes, roles, and example outputs within Carbon Design System constraints
Workflow detail modal: overview, target roles, AI capabilities, and example output templates
Expanded idea view with summary, pain point tags, competitors, partners, recommended actions, and refinement history
Idea detail: expanded summary, color-coded pain points, competitive landscape, partners, and refinement history
Empty state for adding data sources with illustration
Data sources added and ready for processing with tagged metadata
Step-by-step data ingestion: empty state with clear actions, then tagged data sources ready for AI processing
Generated Ideas grid showing AI-produced product concepts with summaries
AI-generated product ideas with summary cards, selection for combining, and chat-to-refine actions
Idea Refinement Assistant with Watson conversational AI
Conversational refinement: Watson assistant helps users iterate on generated ideas with contextual reasoning
Custom workflow builder with kanban-style phase columns
Custom workflow builder: drag AI-enabled steps across product lifecycle phases

Design decisions that shaped the product

01
Role-first architecture
Instead of showing every capability, the dashboard filters by who you are and what phase you're in. A Product Strategist in Exploration sees a completely different set of recommendations than a Quality Engineer in Realization. This single decision eliminated the need for training documentation.
02
Progressive disclosure over power-user defaults
The old platform showed configuration upfront. The redesign uses a guided flow with sensible defaults. Each step is a clear action: add data, process, review, refine, export. Advanced options exist but don't block the path.
03
AI as labeled collaborator
Every AI-powered step is explicitly tagged. The Watson chat in Idea Refinement doesn't just generate output. It explains its reasoning and asks clarifying questions. Users understand what the AI is doing and why, building trust instead of suspicion.
04
Consistent left-rail navigation
Whether you're in Data Scraping or Market Fit, the left sidebar shows your full journey with completion states. Users never lose context about where they are in the workflow or what's already been completed.
05
Templates as starting points, not cages
The workflow builder lets teams compose custom pipelines, but every user starts with a pre-built template matched to their role. This respects both the first-time user who needs structure and the power user who needs flexibility.

What changed

The redesigned IGX platform shipped within the 16-week engagement. The step-by-step experience replaced the previous configuration-heavy interface, and teams were able to run AI-driven product research workflows independently for the first time.

The role-based filtering system reduced the number of visible options on the landing screen by over 60%, depending on the user's role and phase selection. The guided flow structure eliminated the need for the facilitated training sessions that were previously required for every new team onboarding to the platform.

Next project
Johnson & Johnson