Align Before Use
Before B2 uses sensitive client information with external AI capabilities, B2 aligns with the client on the permitted environment and use.
Trust & Information Practices
Modern consulting methods require modern information discipline.
B2 uses AI and other modern tools where they improve research, analysis, development, documentation, and knowledge work without sacrificing client trust, responsible judgment, or control of sensitive information.
Operating principles
AI-Augmented. Human Accountable.
Use the capability. Understand the limits. Protect the client context. Keep people accountable.
Minimize Exposure
B2 uses only the information necessary for the task where practical and avoids unnecessary downstream access, retention, or reuse.
Use Approved Environments
Client requirements and information sensitivity determine which tools and environments B2 uses. Convenience does not decide the environment by itself.
Apply Least Privilege
People, applications, agents, and connectors should receive only the access necessary to perform their intended work.
Keep Humans Accountable
AI can assist with research, analysis, drafting, software development, and review. A person remains responsible for interpretation, validation, recommendations, decisions, and client deliverables.
Validate in Proportion to Risk
B2 scales review and validation to the consequences of error rather than treating every AI-assisted output the same.
Preserve Traceability Where It Matters
Some work requires clear records of sources, decisions, changes, approvals, and assumptions. B2 maintains that traceability when the engagement risk and client environment require it.
Close Out Deliberately
At the end of an engagement, B2 addresses access, credentials, shared information, temporary environments, retained materials, and client requirements.
Sensitive information
“Not used for training” does not answer every stewardship question.
Appropriate use can also depend on data classification, account environment, retention, connectors, agent permissions, identity and access controls, logging, review, and contractual commitments.
- Classify information before sensitive AI use
- Align with the client on permitted tools and environments
- Use least-privilege access for people, agents, and connectors
- Keep human review proportional to risk
- Address retention and access again at engagement closeout
B2 Solutions Lab