Virtual Engagement Officer Overview

Autonomous AI Overview

The Virtual Engagement Officer (VEO) uses autonomous AI to independently manage donor relationships through a series of 8–12 touchpoints annually, guiding donors toward the natural outcome of giving.

The VEO operates using two distinct, synchronized components that work together to make decisions and communicate with donors without requiring ongoing human input.

Moves Management Model

Core Function

The Moves Management Model is a fully autonomous AI decision engine that analyzes donor portfolios to determine the precise next action in the fundraising moves management cycle.

It is trained on fundraising best practices and uses synthetic data generation and predictive modeling to evaluate:

Donor relationship stage
Engagement history with the VEO
Giving patterns and readiness indicators

Based on this analysis, the model identifies the optimal move and timing to advance each donor relationship strategically.

This engine operates independently and does not require human input for decision-making.

Move Categories

Introduction
Cultivation
Ask
Ask More
Stewardship

Sub-Purpose Options

Engage
Invite
Educate
Introduce
Solicit

Decision Inputs

The Moves Management Model evaluates decisions using:

Donor data (organization-provided data plus publicly available data, such as geographic wealth indicators)
Previous interactions and engagement history
Donor relationship stage and readiness indicators

Communication Method Selection

Once a move is selected, the AI determines the optimal delivery channel:

Email
Text message
Robotic handwritten note

Content Generation (LLM)

Core Function

The VEO’s Large Language Model (LLM) serves as the content creation engine. It translates strategic decisions from the Moves Management Model into personalized donor communications.

The LLM synthesizes:

The selected move and sub-purpose
Comprehensive donor intelligence
Historical interaction context
Organization-specific guidelines

It then generates tailored content that aligns with the organization’s voice and deploys it directly through the selected communication channel.

Prompt Formulation Process

The LLM builds prompts using:

Selected move and sub-purpose

Organization-provided donor data
- All donor PII is anonymized when passing through the LLM
- Personalization is added back before deployment using data tags
  
  Historical interaction context
  
  System settings and constraints, including:
- Guardrails (content boundaries, e.g., “do not discuss genetic research”)
- Portfolio context (strategic direction, priorities, and prohibited actions, e.g., “prioritize volunteer opportunities,” “promote the April Gala when relevant,” “use only ABCD, not ABC”)

Content Development

The LLM:

Accesses the organization’s secure knowledge base
Generates personalized message content
Deploys the message directly to the donor via the selected channel

Organization Knowledge Base

The VEO’s LLM accesses a secure, organization-specific knowledge base to generate content. This knowledge base is built through:

Seeding the VEO with marketing communications such as solicitations, event invitations, impact stories, and newsletters
Regular scraping and crawling of approved organization websites
Social media integrations from approved organization social media handles

The VEO can only reference organization-generated content. For example, if the organization appears in the news, the VEO can only share that information after it has been published on one of the organization’s official channels.

Key Safeguards

Human-in-the-loop verification
- VEOs can receive donor replies and are often designed to elicit responses
- All VEO responses in two-way engagement are reviewed by the Version2 Operations Team before deployment
- Reviews verify accuracy, source integrity, and appropriate human handoff recommendations
  
  Guardrails
- Systematic constraints defining messaging boundaries
  
  Portfolio Context
- Strategic parameters that guide VEO focus areas and limitations