← Vault Index
Source: frameworks/kit-candidate-experience-journey/07-process-agent.md

Process Agent — AI-Assisted Candidate Experience Workflow

What This File Covers

This file defines where and how AI tools can be used in the candidate experience journey production and execution. It supplements the consultant process — it does not replace it. Every AI-generated communication requires practitioner review before it goes to a candidate.

The agent process has two phases:

  1. Design phase — AI assists in building the journey (drafting communication templates, assembling candidate packages, structuring the journey map)
  2. Execution phase — AI assists in running the journey (personalizing communications, tracking cadence, drafting rejection and warm communications in real time)

Design Phase — Building the Journey

Drafting Communication Templates

When to use AI: After the extraction interview is complete and the stage-by-stage touchpoints are mapped. AI drafts communication templates for the practitioner to review and refine.

What to provide:

Expected output: Draft templates with personalization fields marked. The practitioner reviews, edits for voice and accuracy, and approves.

Quality check:

Common AI failure modes:

Assembling Candidate Packages

When to use AI: After the package contents are defined (interviewer information, schedule, presentation instructions, organizational materials). AI assembles the formatted document.

What to provide:

Expected output: A formatted candidate package ready for practitioner review.

Quality check: Every fact verified — interviewer names spelled correctly, schedule accurate, virtual links working, presentation instructions complete.

Structuring the Journey Map

When to use AI: After extraction notes are organized by stage. AI produces a formatted journey map document.

What to provide: The complete stage-by-stage touchpoint notes from extraction.

Expected output: A structured journey map with all touchpoints, triggers, timing, methods, and owners organized in a consistent format.

Quality check: Completeness — does the map cover every stage transition? Are there gaps where a candidate could enter a transition with no defined communication?


Execution Phase — Running the Journey

Personalizing Communications from Templates

When to use AI: When the practitioner needs to send a communication from a template with candidate-specific personalization.

What to provide:

Expected output: A ready-to-send communication.

Quality check: The practitioner reads every AI-personalized communication before sending. AI personalization is a first draft, not a final product.

Cadence Tracking

When to use AI: Ongoing during the search. AI monitors when warm communications are due based on the cadence commitments in the journey.

What to provide:

Expected output: Alerts when a warm communication is due for a specific candidate.

Quality check: The practitioner decides what to send and when. AI flags the timing; the practitioner owns the response.

Bulk Close-Out Communications

When to use AI: When the role is filled and close-out communications need to go to all remaining candidates.

What to provide:

Expected output: Personalized close-out communications for each candidate, with language varying based on how deep they went in the process.

Quality check: The practitioner reviews each communication before sending. A candidate who made it to finals gets a different close-out than a candidate who was screened and held. AI can draft the variations; the practitioner confirms each one is appropriate.

Rejection Communication Drafting

When to use AI: When the practitioner needs to deliver a rejection and wants talking points (for phone) or a draft email.

What to provide:

Expected output: Phone talking points or a draft email.

Quality check: Rejection communications are sensitive. AI drafts; the practitioner evaluates whether the tone, specificity, and framing are right for this specific candidate in this specific moment.


Where AI Cannot Replace the Practitioner

Phone conversations. Rejection calls, offer calls, negotiation calls, and any live conversation with a candidate. AI can prepare talking points; the practitioner has the conversation.

Tone judgment for high-stakes communications. When a finalist is rejected, when a candidate is upset about timeline, when an offer is declined — the communication must be calibrated to the specific situation. AI produces drafts; the practitioner decides whether the draft is right.

Accommodation decisions. How to adjust the process for a candidate who needs accommodations requires situational judgment, legal awareness, and organizational flexibility.

Relationship management. The candidate's experience is fundamentally relational. AI produces communications; the practitioner maintains the relationship.

Ethical judgment. When a candidate reveals something in confidence, when there's a conflict of interest, when a communication could be misinterpreted — these require human judgment.


AI Tool Selection

This kit does not prescribe a specific AI tool. Considerations:


What AI Does Not Do

Send communications without practitioner review. Every candidate-facing communication goes through the practitioner before it goes to the candidate. No exceptions.

Make disposition decisions. AI can track where candidates are in the process. It cannot decide who advances, who is held, or who is rejected.

Replace warmth with efficiency. The purpose of AI in the candidate experience is to help the practitioner deliver a better, more consistent experience — not to automate the humanity out of it. If AI-assisted communications feel less personal than what the practitioner would write unassisted, the AI is making the experience worse, not better.