Volume III
Technology Addendum
Written instructions for AI work.
AI work follows the written review and approval process defined in
Volumes I and II.
Volume III- Technology Addendum
Volume III documents written AI instructions that follow the department’s review and approval structure defined in Volumes I and II.
This volume defines how AI may support recurring department work while preserving the
existing review and approval structure.
The addendum does not introduce software.
It does not implement technology.
It does not redesign internal workflow.
It documents:
• where AI may assist preparation
• where human review remains required
• where final approval remains with leadership
• where escalation thresholds apply
Ownership and approval remain unchanged.
This volume documents written instruction, not technology implementation.
How Leadership Uses
this Addendum
For recurring work, the addendum documents written AI instructions that:
• specify what information AI may review
• define the tone and format of responses
• reinforce required review steps before work leaves the firm
• follow escalation thresholds defined in Volume II
AI work is prepared within the department’s written review
process.
What This Documents
The addendum records how AI may operate within the department’s existing review and approval structure.
It documents:
• who prepares the work
• who reviews the work
• who provides final approval
• where questions escalate
• where human judgment remains required
AI may assist preparation.
Leadership continues to review and approve what leaves the firm.
AI Instruction Architecture Levels
Volume III documents written AI instructions that follow the department’s review structure defined in Volumes I and II.
Firms may extend these instructions depending on the type of recurring work where AI use is appropriate. Each level increases the preparation tasks AI may support while review and approval remain unchanged.
Volume III Instruction Packages
Firms select the governance tier where written AI instructions will be developed.
Each governance tier expands the scope of written AI instructions while review and approval remain unchanged.
Tier I: Analysis
Supports preparation such as:
• checking monthly reports before leadership review
• reconciling conflicting numbers across teams
• investigating unusual changes in reports
• explaining changes in results before meetings
• preparing internal analysis summaries
Tier 2: Communications
Includes Tier I and expands support to preparation of:
• executive briefing documents
• presentation materials for leadership meetings
• client proposals or summaries
• internal update communications
• communications plans for initiatives
• workshop or training materials
Tier 3: Leadership Decisions
Includes Tiers I–II and supports preparation for:
• strategy update materials
• evaluating new opportunities
• reviewing active initiatives
• comparing strategic options
• preparing board or executive discussions
Strategic Instruction Requests
For complex or high-stakes work, leadership may request additional written AI instructions aligned to the review structure documented in Volumes I and II.
Leadership provides the context of the question or project.
These instructions may support:
• evaluation of complex information
• structured reasoning
• comparison of strategic options
• preparation of leadership briefings
All instructions follow the review and approval structure documented in Volumes I and II.
Leadership remains responsible for final judgment.
The Concierge Instruction Set™ is a written instructional document prepared for leadership review. It does not provide software, system integration, or automation tools.
Authority Before Automation
Generative AI tools work best when leadership first defines
how recurring work is reviewed before it leaves the firm.
Volumes I and II document how review and approval currently
occur within a department.
Once that structure is written, AI tools may assist in preparing
work without changing who reviews it or who approves it.
Begin by documenting the structural baseline.