R2 Labs
Applied Research and Engineering for AI-Native Work
R2 Labs is the applied research and engineering engine behind R2. It develops open standards, reference architectures, prototypes, and reusable intellectual property that inform R2 Advisory's engagements and R2's emerging products.
Mission
Advisory that is not informed by real engineering work tends to produce recommendations that do not survive contact with production. R2 Labs exists to close that gap.
R2 Labs is a focused applied-research and engineering practice. Its work takes the form of working specifications, reference architectures, prototypes, and experimental systems that R2 Advisory draws on directly in client engagements, and that in some cases mature into R2’s own emerging products.
Research Themes
Six areas of active inquiry.
Delegated Authority and Agent Governance
Defining what an AI agent is authorized to decide, do, access, delegate, escalate, and prove: the foundation of the Agent Delegation Contract.
AI-Native Operating Systems
Reference architectures for coordinating people, agents, models, tools, and authority inside a single operating environment.
Enterprise Memory and Knowledge Work
How organizations preserve context, judgment, and decisions across agents, sessions, and time, not just documents.
Secure Mission AI
Applying architectural discipline from operational technology and federal mission systems to AI that acts in high-consequence environments.
Human-Machine Coordination
Working models for how people and AI agents share authority, escalation paths, and accountability in day-to-day operations.
Governed Expert Work
How professionals design and direct the AI-enabled systems through which their judgment is expressed, scaled, and evaluated.
Current Initiatives
What R2 Labs is building now.
Agent Delegation Contract
An open, portable, machine-readable specification for defining delegated agent authority. Runtime-neutral, model-neutral, and vendor-neutral.
View the ADC specification →iDragonFly
An emerging commercial operating environment for governed expert work, applying R2 Labs research on authority, memory, and coordination.
Explore iDragonFly →Collaboration Model
R2 Labs research is open where it can be. The Agent Delegation Contract is published as a runtime-neutral, model-neutral, vendor-neutral specification so it can be evaluated, adopted, or extended independent of any single vendor. Where research feeds a commercial product, such as iDragonFly, R2 states maturity and status plainly.
Organizations exploring a design partnership, pilot, or collaboration on any Labs initiative are welcome to reach out directly.
Explore a collaboration with R2 Labs.
Whether it is the Agent Delegation Contract, iDragonFly, or a research theme not yet public, we would welcome the conversation.
Discuss a Priority