📄 Something Somewhere SOP for Odoo Implementation Projects

Something Somewhere SOP for Odoo Implementation Projects (Odoo 19)

1. Purpose

To standardize how our team uses AI to:

  • Produce a consistent Solution Overview for implementors and clients

  • Produce a consistent client-facing status update in the task description

  • Set Allocated Time automatically based on the solution’s estimated effort

This improves clarity, reduces manual effort, and standardizes documentation across all projects.

2. Scope

Applies to all implementation projects using Odoo Project tasks where the following fields exist:

  • Standard fields: Title, Stage, Start date, Deadline, Allocated Time, Total Time Spent, Progress, Dependencies, Subtasks

  • Custom fields (as implemented in our SOP design):

    • User Story

    • Acceptance Criteria

    • Changelog Summary

    • Solution (HTML)

    • Documentation (notes)

    • Flags (Standard?, In Development, In Staging, In Production, For Documentation)

    • eLearning Content (reference)

    • Knowledge Article (reference)

3. Definitions

  • Solution (HTML field): A structured, readable solution overview, including an Estimated effort (hours) range.

  • Description: Client-facing status update (Markdown), optionally appending a customer-friendly HTML Solution Overview.

  • Allocated Time: Planned effort in hours for the task. Set automatically to the MAX value in the estimate range (e.g., “8–12” → 12).

4. Roles & Responsibilities

  • Functional Consultant / Implementor

    • Writes/validates User Story and Acceptance Criteria

    • Runs AI actions and reviews outputs for correctness

  • Project Manager

    • Ensures task status updates are client-ready

    • Ensures Allocated Time aligns with scope

  • System Admin

    • Maintains AI prompts, automation rules, permissions, and field mapping

  • QA / Reviewer (optional)

    • Spot-checks output quality and consistency on key milestones

5. Governance & Quality Rules (Mandatory)

  1. AI output is a draft: A human must review before client exposure.

  2. Do not add confidential data into prompts/fields that will be client-visible.

  3. Do not invent information: AI must only use task fields and linked references.

  4. If User Story is blank, AI output must be empty (no auto-fill).

  5. Use standard-first solutions; only propose customization when necessary.