Assistant
Inform Online – Institutional Memory Solutions Through Search and Social Feature Innovations (Northrop Grumman)
Session from Inform Online 2020
How One of the World’s Largest Defense Contractors Turned Decades of Legacy Engineering Data into Accessible, Searchable Institutional Intelligence
Northrop Grumman is one of the world’s leading global defense and aerospace technology companies — a $40B+ revenue organization that designs, builds, and supports some of the most complex engineering programs in existence, including advanced aircraft, space systems, missile defense, and cybersecurity platforms. The company employs tens of thousands of engineers, scientists, and technologists across programs that can span decades and involve classified information, multi-generational design decisions, and engineering knowledge that is irreplaceable once the people who hold it retire.
The institutional memory challenge at Northrop Grumman is not an abstract knowledge management problem. It is an operational and mission-critical risk: when a senior engineer who has worked on a platform for twenty years retires, what happens to the knowledge that exists in their head, in their historical design files, in the project documentation of programs that ended years before current team members joined? When a new engineer needs to understand why a design decision was made a decade ago, or find precedent from a related program across a different business unit with different security clearances, how do they access information that technically exists but is practically inaccessible?
This session shares how they addressed this challenge, unifying siloed legacy data sources across the organization into a platform that makes institutional knowledge discoverable, connects current engineers to the expertise embedded in historical records, and uses AI-powered search and knowledge graph capabilities to surface the right information to the right people regardless of where in the organization it resides.
Watch the video
What the Session Covers
The Institutional Memory Problem in Defense and Aerospace
Defense and aerospace engineering programs generate and depend on institutional knowledge at a scale few other industries match: program histories spanning decades, design decision rationale documented across thousands of files in multiple legacy systems, classified and unclassified knowledge separated by security boundaries, and the expertise of engineers who may be the only people in the organization who understand why specific decisions were made on long-running programs. The session describes how this institutional memory fragmentation manifests operationally at Northrop Grumman, the specific friction points that occur when engineers cannot find historical knowledge they need, and the operational and program risk that creates.
Unifying Siloed Legacy Data: The Architecture Challenge
Northrop Grumman’s data environment reflects decades of engineering program evolution: multiple legacy systems, varied document formats, knowledge spanning both classified and unclassified domains, and data that has accumulated across organizational units that did not share a common information architecture. The session covers how Sinequa’s enterprise search platform connected across these disparate data sources providing a unified search layer that makes institutional knowledge discoverable without requiring the organization to migrate or consolidate its legacy data environment. The connector-based architecture that preserves existing data governance while making content universally searchable is directly relevant to other defense and aerospace organizations facing the same legacy data challenge.
Expert Discovery: Finding Who Knows What Across a Distributed Engineering Organization
One of the most valuable capabilities Northrop Grumman built into their knowledge platform is expertise identification, the ability to surface not just documents but people: engineers and scientists who have worked on related programs, contributed to relevant research, or documented expertise in the technical area a current team needs. The session demonstrates how AI-powered expertise profiling turns the implicit knowledge embedded in historical documents, project records, technical reports, design documentation into an active map of organizational expertise that current teams can navigate. This is the functional equivalent of asking “who in this organization knows the most about this specific technical problem” and getting an answer grounded in the actual work record of every employee, not just self-reported profiles.
Knowledge Sharing and Collaboration Features: Connecting the Organization’s Knowledge Graph
The session covers how Northrop Grumman enhanced the core search platform with social and collaborative knowledge features that make institutional memory a living, actively maintained resource rather than a static archive. This includes how employees can contribute to, validate, and build on the knowledge environment creating a feedback loop between current engineering work and the historical knowledge base that prevents institutional memory from becoming stale or incomplete as programs evolve. The “LinkedIn-style” social features referenced in the original framing represent something more specific in a defense context: a structured expertise network that allows engineers to find peers with relevant experience, even across organizational boundaries and security domains, without exposing classified program details.
Security Architecture: Making Knowledge Accessible Without Making It Accessible to Everyone
Northrop Grumman operates across classified and unclassified domains with security clearance requirements that are among the most rigorous in any industry. A knowledge platform that does not enforce these boundaries that allows a query from an uncleared engineer to surface content from a classified program, is not a knowledge platform; it is a security incident waiting to happen. The session addresses how Sinequa’s access control architecture enforces security classification boundaries at the retrieval layer, ensuring that the unified search experience respects the security architecture of the underlying data environment without requiring engineers to manually navigate those boundaries themselves.
Frequently Asked Question
Institutional memory is the accumulated knowledge, experience, and expertise that an organization has built over time — the understanding of why decisions were made, how programs evolved, what was tried and failed, and where the specialized expertise resides that current work depends on. In defense and aerospace, institutional memory is operationally critical for specific reasons that do not apply to most other industries. First, program lifecycles are extremely long: aircraft, missile systems, and space platforms are designed, built, and operated over decades, meaning current engineers routinely need access to design rationale and technical decisions made before they joined the organization. Second, workforce demographics create knowledge transfer risk: many defense organizations have significant portions of their expert engineering workforce approaching retirement, carrying program knowledge that is genuinely irreplaceable if it is not captured and made accessible. Third, security classification creates knowledge fragmentation: classified program knowledge cannot be searched alongside unclassified knowledge without careful access control architecture, creating information silos that exist by design but impose real operational costs on engineers who need cross-program context.
Enterprise AI search addresses institutional memory loss through three interconnected mechanisms. First, by indexing the implicit knowledge embedded in historical documents — design records, technical reports, project documentation, email archives, and the accumulated output of years of engineering work — and making it searchable through natural language queries, the platform preserves institutional knowledge even after the people who created it have moved on. Second, by building expertise profiles from that document record, the platform identifies who in the current organization has worked on related problems, contributed to relevant research, or documented experience in specific technical domains — connecting current teams to living expertise they might not know exists. Third, by connecting across the organization’s full data environment regardless of which legacy system content was created in, the platform makes historically siloed institutional knowledge discoverable through a single search experience. Northrop Grumman’s deployment demonstrates all three mechanisms applied to one of the most demanding knowledge management environments in industry.
Security classification in defense environments requires an access control architecture that operates at a fundamentally different level of rigor than standard enterprise access management. Sinequa’s early-binding security model enforces access permissions at the retrieval layer — at the moment a query is processed, the system checks the requesting user’s authorization for every data source against which the query would retrieve content. This means that an engineer without clearance for a specific program cannot receive search results that draw on classified content from that program, even if they are using the same search interface as cleared colleagues. The access control is applied consistently across every connected data source, based on the permission model of the originating system, without requiring manual security configuration in the search platform itself. This architecture makes it possible to provide a unified search experience across classified and unclassified knowledge environments while maintaining the hard boundaries that security classifications require — giving engineers the benefit of broad institutional knowledge access within the scope of what they are authorized to see.
Defense and aerospace knowledge management differs from other enterprise knowledge management contexts in four significant ways. First, program longevity: defense programs regularly span 20–40+ years, meaning knowledge management systems must maintain accessibility across technological generations, organizational changes, and workforce turnover at a timescale few other industries require. Second, security architecture: the mandatory separation between classification levels creates knowledge silos by design, requiring search and AI systems that can operate across these boundaries while strictly enforcing them — a technical requirement that most enterprise search platforms were not designed to meet. Third, documentation density: defense programs generate extraordinary volumes of technical documentation — specifications, test reports, design analyses, regulatory submissions, program reviews — that represents accumulated institutional knowledge of immense value but is practically inaccessible without sophisticated search and retrieval. Fourth, workforce demographics: defense engineering workforces have faced sustained retirement waves among experienced personnel, creating acute knowledge transfer urgency that makes institutional memory preservation a strategic priority rather than a background IT concern.
Share
