Process Agent — AI-Assisted Referral Program Workflow

What This File Covers

This file defines where and how AI tools can be used in referral program design and execution. Every AI-generated output requires practitioner review before it reaches the client or any audience.

Design Phase

Drafting Referral Communications

When to use AI: After audience segmentation is complete and the position summary exists. AI drafts communications for each audience.

What to provide:

• Audience identity (staff, board, external, etc.)
• The position summary
• Tone guidance per audience
• Incentive details (if applicable and approved)
• Confidentiality constraints
• The practitioner's natural framing from extraction notes

Expected output: Draft communications per audience, ready for practitioner review.

Quality check:

• Specific enough that referrers can identify appropriate candidates?
• Differentiated by audience (not the same email with a different salutation)?
• Incentive details accurate and approved?
• Confidentiality language present where needed?
• Sounds like the practitioner, not like a recruiting blog?

Common AI failure modes:

• Generic ask ("if you know anyone interested") instead of specific criteria
• Same communication for all audiences with cosmetic changes
• Including incentive details that haven't been approved
• Overly formal tone for staff communications, or overly casual tone for board communications
• Missing confidentiality language when the search is confidential

Drafting Social Media Content

When to use AI: After the position summary exists and the social media component is confirmed. AI drafts announcement content.

What to provide:

• Position summary
• Client's social media voice (from marketing/communications)
• Publication channels
• What action the reader should take (apply, refer, share)

Expected output: Draft posts for the client's marketing team to review and publish.

Quality check: The practitioner does not post — the client's marketing/communications team reviews, approves, and publishes.

Position Summary for Referrers

When to use AI: After the position profile is finalized. AI adapts the profile into a shareable summary for referrers.

What to provide:

• The full position profile
• The target referrer audience (what do they need to know to identify good candidates?)
• Length constraint (referral summaries should be significantly shorter than position profiles)

Expected output: A concise summary that gives referrers enough to identify appropriate candidates without overwhelming them.

Quality check: Does it hit the must-haves without listing every qualification? Would a referrer reading this think of a specific person they know?

Execution Phase

Referral Acknowledgment Drafting

When to use AI: As referrals come in. AI drafts personalized acknowledgment communications.

What to provide:

• Referrer name
• Referred candidate name
• Acknowledgment template
• Any personalization context

Expected output: Ready-to-send acknowledgment. Practitioner reviews each before sending.

Referrer Update Drafting

When to use AI: At defined milestones (candidate contacted, search concluded). AI drafts update communications.

What to provide:

• Referrer name
• Milestone reached
• Update template
• Confidentiality boundaries (what can and can't be shared)

Expected output: Draft update. Practitioner reviews for accuracy and confidentiality compliance.

Referral Quality Analysis

When to use AI: After referrals are received. AI compares referred candidates against the position profile's must-haves.

What to provide:

• Referral information (name, background, whatever the referrer provided)
• Position profile must-haves

Expected output: A preliminary assessment of which referrals appear to align with must-haves and which may not. The practitioner makes the final determination — AI flags, humans decide.

Follow-Up Campaign Drafting

When to use AI: At the 2-3 week mark if the referral pipeline needs refreshing. AI drafts a follow-up communication that updates referrers on progress and renews the ask.

What to provide:

• Original referral ask
• Progress update (how many referrals received, whether more are needed, any refinement to what kind of candidates would be most helpful)
• Tone guidance

Expected output: A follow-up communication that feels like a genuine update, not a repeated ask.

Where AI Cannot Replace the Practitioner

Audience identification and sequencing. Who to ask, in what order, and what political dynamics to navigate.

Incentive design and authorization. Financial commitments require human judgment and client approval.

Referrer relationship management. When a board member's referral isn't qualified, when a referrer pushes for information they can't have, when organizational politics intersect with the referral program.

Confidentiality decisions. What can be shared, with whom, and when.

Client approval. The program is deployed only after human review and explicit client authorization.

What AI Does Not Do

Deploy communications without practitioner review. Every communication — referral ask, acknowledgment, update, social media — goes through the practitioner before it reaches any audience.

Authorize incentives. AI can draft the incentive proposal. Approval is human.

Evaluate referral quality as a final determination. AI can flag alignment or misalignment. The practitioner and sourcer make the screening decision.

Replace the personal ask. Some referral asks — particularly to board members or senior stakeholders — are better delivered personally by the practitioner or by the client's leadership. AI can prepare talking points. The conversation is human.