Email Personalization Agent

Adam's take

AI email tools promise magic and deliver spam. This one drafts, a human sends. Keep it that way until your quality scores earn something more.

Required tools

Optional tools

Expected value

Cuts drafting time per personalized email from ~15 min to ~2 min review

Deployable workflow pack

The full pack includes the agent spec, deployment guide, architecture diagrams, n8n workflow skeleton, Clay template, HubSpot properties, JSON schemas, OpenAPI spec, cost and ROI calculators, demo data, QA checklist, and governance notes.

How to deploy this agent

"Personalized" outbound usually means a first-name merge tag. Real personalization — buying role, account signal, the right proof point for that persona — takes a rep ten-plus minutes per email, which is why it only happens for the first twenty accounts on any list. This agent drafts that level of personalization at scale. It does not send anything.

Core signals

Input schema

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://deploytheagent.com/schemas/agent-006-input.schema.json",
  "title": "Agent #006 Email Personalization Agent Input",
  "type": "object",
  "required": [
    "record_id"
  ],
  "properties": {
    "record_id": {
      "type": "string",
      "description": "CRM record ID (company/contact/deal depending on agent)."
    },
    "domain": {
      "type": "string"
    },
    "config_version": {
      "type": "string",
      "description": "Versioned config used for this run."
    }
  },
  "additionalProperties": true
}

Output schema

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://deploytheagent.com/schemas/agent-006-output.schema.json",
  "title": "Agent #006 Email Personalization Agent Output",
  "type": "object",
  "required": [
    "record_id",
    "personalization_quality_score",
    "email_compliance_status",
    "email_draft_status"
  ],
  "properties": {
    "record_id": {
      "type": "string"
    },
    "personalization_quality_score": {
      "type": "string",
      "description": "Agent #006 output field."
    },
    "email_compliance_status": {
      "type": "string",
      "description": "Agent #006 output field."
    },
    "email_draft_status": {
      "type": "string",
      "description": "Agent #006 output field."
    }
  },
  "additionalProperties": false
}

Scoring / classification logic

A draft has to pass both the quality threshold and the compliance check before it enters the human review queue at all. Drafts that fail either check route back for regeneration or manual flagging — they never reach a rep’s inbox as a ready-to-send option.

HubSpot CRM properties

n8n workflow skeleton

The Email Personalization Agent drafts persona-aware outbound emails using prospect context, buying role, account signals, and approved proof points, then scores each draft for quality and compliance before a human ever sees it. It does not send anything. Configure every node marked CONFIGURE before enabling the trigger.

1. Trigger
   CONFIGURE: set cadence per agent-spec.md's What It Does section. Leave disabled until Step 6 of deployment-guide.md.
2. Load versioned config
   CONFIGURE: replace with your versioned config built from real data (see deployment-guide.md Step 1).
3. Fetch source data
   CONFIGURE: {{HUBSPOT_CREDENTIAL}}. Pull the fields listed in agent-spec.md's Inputs table.
4. Validate + gate records
   Drops records missing required inputs or failing suppression/gate checks defined in agent-spec.md. Logs every drop.
5. Apply workflow logic
   Implements the Workflow Logic table from agent-spec.md. Writes model/config version alongside every derived value for auditability.
6. Write agent-owned CRM fields
   CONFIGURE: {{HUBSPOT_CREDENTIAL}}. Writes only: personalization_quality_score, email_compliance_status, email_draft_status. Never writes sales-owned fields.
7. Route to human review / alert
   CONFIGURE: {{SLACK_CHANNEL}}. Fires for records requiring human review per agent-spec.md's Governance section.

Cost calculator

ROI calculator

Governance notes

• This agent never sends email — no send permission, no send-trigger access, in any configuration. Every draft requires human review and approval before it enters a live sequence.
• Only approved, versioned proof points from the sanctioned library are usable — the agent cannot invent or reword a proof point on its own.
• The compliance check runs on every single draft, with no exceptions and no bypass path.
• Send-capable credentials are never attached to this workflow — verified explicitly in QA, not assumed.

Failure modes

• Draft gets auto-sent: prevented structurally — the n8n workflow uses read/draft-only sequencing-tool API scopes, and QA confirms no send-capable credential is ever attached.
• A fabricated stat slips through: every miss expands the versioned banned-claims list and is logged as a governance incident, not quietly patched and forgotten.
• Same generic draft regardless of persona: a QA test generates one draft per persona and asserts they meaningfully differ — if they don’t, the angle-selection config isn’t actually branching.
• Proof point used out of context: proof-point library entries require industry and persona tags before they’re eligible for selection at all.

QA checklist

FAQ

Related agents