Case Study / GE Aerospace
Designing the AI-native knowledge operating system for one of the world’s most complex aerospace organizations
AI-assisted learning, contextual retrieval, guided workflows, content governance, and operational guidance, designed inside one of the world’s most complex aerospace organizations.
Work / GE Aerospace
Sectors
Aerospace & Defense
Enterprise Knowledge
Learning & AI
Client
GE Aerospace
New York
Engagement
Existing partner
Phased, low-risk expansion
Capability surface
Intelligence Architecture
AI Learning Systems
Knowledge Orchestration
Content Governance
UX & IA
Design System
Vetted Contractor Scale
FlightDeck is GE Aerospace’s internal operating system for how the company runs. Standards, behaviors, daily management, problem solving, the practices that turn a 34,000-person organization into a coordinated one.
Bttr. designed the layer underneath. A knowledge OS structured clearly enough to be assisted by AI, governed at enterprise scale, and useful long before the model arrives.
The Terrain
We get the world this work lives in.
Enterprise knowledge systems don’t live in product labs. They live inside procurement gates, security reviews, tenant migrations, and platform limits. A beautiful prototype is easy. A system that works after all of that is the job.
We’ve already absorbed the constraints. So the conversation can start at what to do, not at how to navigate the building.
About FlightDeck
A pattern we’ve already solved.
FlightDeck is built on three behaviors and eight operational fundamentals, the shared language of how GE Aerospace operates. The content existed. Templates existed. Examples existed. The hub had been collecting them for years.
What was missing was a system that helped people use any of it. Pages multiplied. Templates drifted. Subject matter experts became bottlenecks. New leaders inherited responsibility before they inherited context. It’s the same pattern any large organization hits when content modernization meets AI ambition.
“There is a lot of content, many templates, lots of guidance. On the other hand it is a bit like a jungle, what do you ideally apply for what?”
“Nobody has given me training on how or why to use Flight Deck.”
“Starting to become too crowded with information.”
The Brief
From repository to guidance.
Most enterprise systems are built like libraries. They store information. But operational teams don’t need another library. They need a guide.
A frontline leader doesn’t come into a system thinking about taxonomy. They come in with a job: run a better daily management meeting, choose the right standard work template, coach a team through a problem, find an example from another site, know what good looks like.
FlightDeck had answers. It didn’t always create a path. The brief was to design the path, in a way that survives enterprise reality.
The Hub
A working system, not a deck about one
The deliverable was a live, password-gated stakeholder hub. Eleven surfaces, cross-linked to source data on every claim, designed so leadership could click through the strategy instead of read about it.
How We Engage
Scope expands. Coordination doesn’t.
Bttr. is already inside the operating rhythm at GE Aerospace. New scope drops in cleanly. Capacity flexes through a vetted bench, so the team scales up or holds steady without adding vendors or coordination overhead.
Already inside
No new vendor onboarding. The work begins the week the decision is made.
Bench, on call
Vetted senior contractors across UX, content, design system, and AI strategy. Up or down inside the same engagement.
Single accountability
One account spine across every capability. No vendor coordination overhead falling back on the GE team.
Phased Rollout
Three phases. Each one defensible on its own.
PHASE 01
Stabilize · ~30 days
Architecture & alignment
Audit the content model. Lock the template language. Confirm the priority audiences. Surface the integration mechanism. Acceptance gate: design approved, integration locked.
PHASE 02
Standardize · ~60 days
Build the spine
Ship the redesigned hub, the fundamental pages, the template library, and the operating cadence for governance. AI scoped around four jobs the system can actually support: help, curate, coach, route.
PHASE 03
Expand · ~30 days
Hand the loop back
Hardening, UAT, support window, and a documented monthly operating loop so the GE team owns it cleanly after Bttr. steps back. The system compounds with the factory running.
Each phase ships something the team can show internally. No big-bang release. No quarter-long quiet period before evidence shows up.
The Knowledge OS
Eleven capabilities,
one connected layer.
AI alone doesn’t fix a knowledge problem. The system around it does. Bttr. covers the whole layer, so the team buying it isn’t left to integrate eleven vendors.
Intelligence
01 / 03
- 01
AI Assistant
Trusted-source answering inside the workflow.
- 02
Contextual Retrieval
Knows where the user is and what they’re trying to do.
- 03
Knowledge Surfacing
Right artifact, right moment. No search detour.
- 04
Intelligent Discovery
Examples, peers, experts, surfaced by context.
Workflow
02 / 03
- 01
Guided Workflows
Decisions, not destinations. The path is the product.
- 02
Role-Based Journeys
Different surfaces for leaders, operators, learners.
- 03
Embedded Assistance
Help that lives at the point of work.
- 04
Coaching Layer
Standards turned into guided behavior, not docs.
Foundation
03 / 03
- 01
Content Governance
A monthly operating loop so signal doesn’t decay.
- 02
Standards & Templates
A repeatable shape for every fundamental.
- 03
Learning Orchestration
Onboarding paths and standards woven into use.
AI sits inside the layer, not on top.
That’s what makes the assistance actually useful when it ships, and what makes the content investment compound instead of decay.
Five Concept Directions
One hub, five entry models
Each concept rendered as a real SharePoint hub page with the GE chrome, real persona content, and grounded in voice-of-customer research. Not a deck. Five shippable directions the leadership team could click through.
A · Stage-Based
"Where are you in your journey?"
B · Wingmate AI
"Skip the search. Just ask."
C · Maturity
"Show me where to focus this week."
D · Personas
"Choose your starting point."
E · Guided Flow
"Just tell me what to do next."
The Walkthrough
A guided tour through all five.
Concept C · Maturity Dashboard
A leadership-first front page
Aggregate score against a peer group. Ranked opportunities tied to the Hoshin plan. Site leaderboard with real GE plant data. The hub becomes a place to make a decision, not a place to find a file.
Inside the AI Layer
Four concrete jobs
an assistant can take on.
Not a chatbot. A working assistant that lives inside the workflow, answers from trusted source material, and routes the user to the next right action.
Mode 01 · Help
Answer from trusted content.
Anchored in 30+ tracked pages, top-20 templates, and the operating-moment tags applied at Station 03 of the comment factory.
User
“”
Assistant
Standard Daily Mgmt Board, the most-pulled template in the hub at 14,572 views. Tagged for manufacturing and transactional teams. Three real examples sit one click below.
Concept E · Guided Flow
Just tell me what to do next.
A 3-column wizard with a connecting timeline rail. Designed for the FLIGHT DECK leader who walks in mid-week without time to browse. The interface is the answer.
01, The Architecture
Built around the work, not the org chart.
Every fundamental was rebuilt around a repeatable shape: what is it, when to use it, what good looks like, what templates support it, what examples exist, who can help. Different subject matter, same operating logic. Easier for humans to navigate. Easier for AI to understand later.
- Workflow as the unit of design, not the page
- One repeatable structure across all eight fundamentals
- Behaviors and fundamentals reframed as routes, not destinations
- Three priority audiences, surfaced from voice-of-customer research
02, Capability Layers
AI is one layer. Not the answer.
The strongest enterprise systems aren’t a chatbot bolted onto a hub. They’re a stack of disciplines that move together. Bttr. covers the stack, so the team buying it doesn’t have to coordinate it.
UX & IA
The path through the system. Workflow over page.
Content Architecture
Modular, tagged, reusable, connected. The substrate AI assistance can actually stand on.
Design System
A reusable visual language compatible with GCC High constraints.
AI Strategy
Scoped around four jobs: help, curate, coach, route. Anchored in trusted source material.
Governance & Ops
Monthly operating loop so the system compounds, not decays, after handoff.
Contractor Scale
Vetted senior bench inside the same engagement. Capacity that flexes without onboarding overhead.
03, Content Architecture
Modular. Tagged. Reusable. Connected.
A content refresh isn’t just rewriting. It’s restructuring. Every piece of FlightDeck content was treated as part of a larger knowledge system, tagged with its fundamental, behavior, use case, role, maturity, and the operating moments where it actually shows up.
The goal was not more content. The goal was better signal. Once the system understood what each asset was, where it belonged, and when it should appear, the entire experience became more useful, for users, for administrators, and for any future AI assistance.
Built On Real Research
302 comments, read by hand.
The voice-of-customer discovery doc is the substrate everything else stands on. 7 personas across 34,760 employees. An 8.1% / 649-respondent survey. Top-20 templates ranked by actual pull data. AI categorization was tried and dropped because the buckets didn’t fit. We read every comment.
How We Worked
Designed around momentum.
01, Listen
Voice of customer, by hand
AI categorization was tried and dropped. The buckets didn’t fit. We read every open-ended comment manually and let the patterns cluster.
02, Architect
Clarity, not waiting
While access moved through procurement, we advanced the architecture. Content model, template language, navigation, permissions, leadership alignment.
03, Activate
Ready when access opened
By the time the systems unlocked, the work wasn’t starting. It was already moving. A technical audit path was queued and ready to run.
The Posture
“We’ve seen versions of this before. There’s a clean way through it.”
The Shift
From repository to guidance.
Before
- Pages organized by content
- Templates scattered across sections
- Examples buried as attachments
- Users required prior knowledge
- AI opportunity undefined
- Governance dependent on manual upkeep
After
- Workflows organized by intent
- Fundamentals structured consistently
- Templates connected to real use cases
- Examples elevated as proof
- AI scoped around help, curate, coach, route
- Governance built into the content model
Results
Clarity before scale.
3 + 8
Behaviors and fundamentals
A repeatable structure across every operational page in the hub.
5
Connected surfaces
Hub, Fundamental Pages, Template Library, Example Repository, AI Assistance.
11
Knowledge OS capabilities
From AI assistant to coaching layer, covered under one engagement.
~34K
Employees in scope
A platform shaped for the operational reality of a global aerospace organization.
FlightDeck isn’t a moonshot. It’s a system that needs to keep working while it gets better. We’ve seen versions of this before. There’s a clean way through it.
Client
GE Aerospace
Location
New York
Sectors
Aerospace & Defense
Enterprise Knowledge
Learning & AI
Bttr. Roles
Product Strategy
UX & IA
Content Architecture
AI Experience Strategy
Design System
Next Case Study
Tiger BioSciences
A complete digital platform for a regenerative medicine leader
We design, build, and run digital systems for the science, aerospace, and biotech industries.