Brand Sentinel Architecture
The Problem
Currently, our content review process is a manual bottleneck. Every piece of copy: from social posts to email blasts: waits 24-48 hours for a Creative Director to check it against the Mailchimp Style Guide. This “Approval Log-jam” slows down deployment, and human reviewers often miss subtle tone violations when they are rushed.
The Solution
Architect an agentic workflow that monitors the designated Google Drive folder. The Sentinel will:
- Retrieve specific Mailchimp “Voice and Tone” rules for the content type from the RAG system.
- Reason through the copy to identify “Hype,” “Jargon,” or “Off-Brand” language. Hype (-10), Jargon (-7), and Off-Brand Tone (-5) to a 100% baseline.
- Calculate a Brand Alignment Score (BAS) by applying weighted deductions to automate routing: 90%+ to “Approved,” <90% to “Rejected,” and <75% to “Triage.”
- Execute a detailed audit report in the #creative-review Slack channel, highlighting specific changes and reasoning.
- Log audit results to the Mailchimp Brand Audit Ledger Google Sheet for a searchable paper trail.
C.O.R.E. Reference Guide
Context
Act as a [Expert Role] responsible for [goal of project]. Your work begins when you are WATCHING FOR [Event Trigger] in [Tool A]. Use [Source of Truth] as your absolute background context.
Operations
Reason through the information using these steps: 1. GATHER the [Data], 2. PROCESS the [Details], 3. DECIDE the [Result]. For each [unit] perform: [Action].
Record
Create a paper trail: Select [Data Points] and LOG in a relevant spot in [Record Type] named [Document Name].
Execution
Finalize the output: If [positive condition(s)] then [positive action(s)]. If [negative condition] then [negative actions(s)]. If [other condition(s)], then [other action(s)].