AI council + intake process (L3)

The steps

1. 01

   Assemble the AI Governance Council

   To begin, you must formalize the 'AI Governance Council' by inviting one representative from RevOps, IT/Security, Legal, Finance, and a high-level GTM leader (VP level). The mission is to centralize decision-making and prevent 'Shadow AI' where departments buy tools individually. Create a recurring 45-minute weekly calendar invitation titled 'AI Intake & Governance Review.' • Owner: RevOps Lead. • Time Estimate: 1 hour for setup; 45 mins weekly. • Setting up: In your calendar tool (Google/Outlook), create the invite and attach a standardized 'Council Charter' document that defines roles (e.g., Security owns data privacy, Finance owns budget approval, RevOps owns technical integration). • Common Pitfall: Inviting too many people. Keep it to these 5 core roles to ensure agility. • Definition of Done: A recurring calendar invite accepted by all 5 stakeholders with a linked folder for meeting minutes and tool documentation.

2. 02

   Launch the Public AI Intake Form

   Create a standardized intake form using Typeform, Jira Service Management, or Google Forms to act as the single entry point for all AI tool requests. The form must capture: Tool Name, Primary User Group, Business Case (ROI), Data Types Processed (PII, PHI, or internal-only), and Integration Requirements. • Owner: IT/RevOps. • Time Estimate: 2-3 hours. • Specific Fields: Add a 'Problem Statement' field,if the requester cannot explain the specific manual task being automated, the request is flagged. Include a dropdown for 'Alternative Tools' to check if they explored existing stack capabilities first. • Common Pitfall: Making the form too long, which encourages employees to bypass the process. Balance rigor with speed. • Definition of Done: The form is live, linked in the company Slack/Teams 'General' channel and the internal Wiki/Notion, with an automated notification (via Zapier or native integration) that alerts the Council when a new submission arrives.

3. 03

   Execute Weekly Review & SLA Tracking

   Establish a Service Level Agreement (SLA) ensuring every request receives an initial 'Go/No-Go' or 'Needs Info' response within 10 business days. During the weekly meeting, use a Kanban board (Trello or Asana) to track requests through the stages: 'Submitted,' 'Security Review,' 'Legal/Privacy Check,' and 'Final Decision.' • Owner: RevOps Lead (as the facilitator). • Time Estimate: 30 minutes per request. • Process: Security checks the tool's SOC2 status; Legal reviews the 'Data Training' clauses (to ensure company data isn't used to train the vendor's models); Finance checks the budget. Use a simple scoring rubric (1-5) for 'Strategic Fit' and 'Risk Level.' • Common Pitfall: Analysis paralysis. If a tool costs <$5k and handles no PII, use an 'Express Lane' for approval to maintain momentum. • Definition of Done: A documented decision for every request in the Kanban board, with a trail of why it was approved or rejected.

4. 04

   Build a Centralized AI Tool Inventory

   Build a 'Source of Truth' inventory to track every approved AI tool across the organization. This should be a shared spreadsheet or Airtable including: Tool Name, License Count, Total Cost, Primary Admin, Renewal Date, and Data Classification. • Owner: Finance/RevOps. • Time Estimate: 4 hours (initial build), 1 hour weekly maintenance. • Fields to Include: 'AI Capability' (e.g., LLM, Image Gen, Transcriber), 'Current Status' (Pilot vs. Production), and 'Review Date.' • Common Pitfall: Forgetting to list 'Implicit AI',features within existing tools (like HubSpot AI or Salesforce Einstein),which can lead to overpaying for redundant standalone tools. • Definition of Done: A master list accessible to all employees (read-only) so they can see which tools are already sanctioned before requesting new ones.

5. 05

   Perform Quarterly Sunset Audits

   Implement a quarterly 'Sunset Review' to prune underutilized or underperforming tools. Export usage logs from the tool's admin panel or your SSO provider (Okta/Azure AD) to determine Monthly Active Users (MAU). If a tool shows <30% MAU or fails to meet its initial ROI goals, it is flagged for decommissioning. • Owner: RevOps/IT. • Time Estimate: 3-5 hours per quarter. • Action: Notify the tool owner 30 days before the renewal date if the tool is at risk. Use a standard email template: 'Your tool [Name] is currently at [X]% usage. Please provide a justification for renewal by [Date] or it will be sunset.' • Common Pitfall: Renewing tools just because 'the team might use it later.' Be ruthless with budget to make room for newer, more effective AI innovations. • Definition of Done: A quarterly report shared with the Council showing which tools were renewed, which were cancelled, and the total cost saved.

Next playbooks

• AI literacy bootcamp for GTM (L2)
• Centralized AI prompt library (L2)
• Claude for Work shared projects (L2)
• Notion AI + Claude for company knowledge ops (L2)