Adoption
Identify the organizational conditions that separate durable deployment from isolated pilots and individual experimentation.
Research area / Enterprise AI
How institutions turn AI capability into operating capacity.
Research on adoption, architecture, operating models, workforce design, economics, controls, and the infrastructure required to move from experimentation to repeatable enterprise deployment.
Adoption is an operating-model problem.
Model access is becoming abundant. Durable advantage increasingly depends on how organizations connect models to data, workflows, permissions, evaluation, security, ownership, and measurable business outcomes.
Architecture
Study how models, retrieval, tools, memory, evaluation, observability, and enterprise data should fit together.
Operating model
Examine which capabilities should be centralized, federated, embedded, or delivered through shared platforms.
Value
Measure productivity, quality, cost, resilience, risk reduction, and strategic optionality rather than model usage alone.
Enterprise AI system
Six layers of institutional readiness.
Our framework connects technical deployment with the governance and organizational capabilities required to operate AI at scale.
Strategy and portfolio design
Prioritize use cases by economic value, risk, data readiness, workflow fit, and organizational ability to absorb change.
Data and knowledge access
Establish governed access to enterprise knowledge, structured data, retrieval systems, and authoritative sources.
Architecture and integration
Design model routing, tool access, memory, observability, evaluation, and human checkpoints as an enterprise system.
Governance and security
Connect inventories, identity, authorization, risk classification, evidence, approvals, and incident response.
Operations and ownership
Define service ownership, support, change management, escalation, monitoring, and lifecycle responsibilities.
Measurement and economics
Track business impact, quality, risk, cost-to-serve, adoption depth, and portfolio-level returns.
Research questions
What we are measuring now.
Where is AI creating durable workflow value?
We distinguish visible experimentation from repeatable institutional capability.
How does agentic AI change enterprise architecture?
We study identity, permissions, orchestration, evidence, and human control as agents gain system access.
Which operating models scale?
We compare centralized, federated, platform-led, and embedded models for AI delivery and governance.
Adoption signals
Measure institutional depth—not AI activity.
Usage is only one signal. We track whether AI is connected to accountable workflows, governed data, repeatable architecture, operating ownership, and measurable outcomes.
Three adoption tests
Workflow depth
Is AI embedded in repeatable work with defined inputs, outputs, owners, and exception paths?
Control depth
Are identity, data access, evaluations, approvals, monitoring, and escalation integrated into deployment?
Economic depth
Can the organization measure quality, cost, cycle time, risk reduction, or revenue impact over time?