AI Project Brief & SOW Template

How to Use This Template

This template helps you define, scope, and gain approval for AI projects. Fill in every [BRACKETED] field with your project's specifics. Share the completed brief with stakeholders for sign-off before beginning work.

Before you start: The most common reason AI projects fail isn't technology — it's unclear objectives. Spend time getting Section 2 (Objectives & Success Criteria) right before worrying about tools and timelines.

Section 1: Executive Summary

Project Name: [PROJECT NAME]

Project Sponsor: [NAME, TITLE]

Project Lead: [NAME, TITLE]

Date: [DATE]

Version: [VERSION NUMBER]

Summary: [2–3 sentences describing the project, the problem it solves, and the expected business impact. Write this last, after completing all other sections.]

Estimated Investment: [TOTAL COST — tools + training + implementation + ongoing]

Expected Return: [QUANTIFIED BENEFIT — time saved, revenue impact, cost reduction]

Timeline: [START DATE] to [END DATE] ([TOTAL WEEKS])

Status: [DRAFT / IN REVIEW / APPROVED]

Section 2: Objectives & Success Criteria

Problem Statement

[Describe the current pain point in 3–5 sentences. Be specific about the impact: hours wasted, errors made, revenue lost, customer complaints, etc. Quantify wherever possible.]

Project Objectives

Primary Objective: [What is the single most important outcome? e.g. "Reduce monthly report generation time from 14 hours to 4 hours."]
Secondary Objective: [e.g. "Improve report accuracy by eliminating manual data entry errors."]
Tertiary Objective: [e.g. "Build internal AI capability and confidence within the finance team."]

Success Criteria

Success Metric 1: [METRIC] — Target: [TARGET] — Measurement Method: [HOW YOU'LL MEASURE]
Success Metric 2: [METRIC] — Target: [TARGET] — Measurement Method: [HOW YOU'LL MEASURE]
Success Metric 3: [METRIC] — Target: [TARGET] — Measurement Method: [HOW YOU'LL MEASURE]

Out of Scope

[Explicitly state what this project does NOT cover. This prevents scope creep.]
[e.g. "This project covers the finance team only. Expansion to other departments will be a separate initiative."]
[e.g. "This project uses existing AI tools. Custom model development is out of scope."]

Section 3: Data Requirements

Data Source	Data Type	Classification	Current Location	Access Required From	Preparation Needed
[e.g. CRM]	[e.g. Customer records]	[Public / Internal / Confidential / Restricted]	[e.g. Salesforce]	[e.g. Sales Ops team]	[e.g. Export, anonymise PII, clean duplicates]
[e.g. ERP]	[e.g. Financial transactions]	[CLASSIFICATION]	[SYSTEM]	[TEAM]	[PREPARATION]
[e.g. Spreadsheets]	[e.g. Manual reports]	[CLASSIFICATION]	[LOCATION]	[TEAM]	[PREPARATION]
[SOURCE]	[TYPE]	[CLASSIFICATION]	[LOCATION]	[TEAM]	[PREPARATION]

Data Privacy Considerations

Does this data contain PII? [YES / NO] — If yes: [ANONYMISATION PLAN]
Data processing agreement required with AI vendor? [YES / NO]
Data residency requirements: [REGIONS / JURISDICTIONS]
Retention and deletion policy: [POLICY]

Section 4: Timeline & Milestones

Phase	Duration	Key Activities	Milestone / Deliverable	Dependencies
Discovery & Planning	[WEEKS]	[ACTIVITIES — e.g. requirements gathering, tool selection, data audit]	[DELIVERABLE — e.g. Approved project plan]	[e.g. Stakeholder availability]
Setup & Configuration	[WEEKS]	[ACTIVITIES — e.g. tool procurement, data preparation, prompt development]	[DELIVERABLE — e.g. Working prototype]	[e.g. IT security approval]
Pilot & Testing	[WEEKS]	[ACTIVITIES — e.g. pilot with 2–3 users, feedback collection, iteration]	[DELIVERABLE — e.g. Pilot results report]	[e.g. Data access granted]
Rollout & Training	[WEEKS]	[ACTIVITIES — e.g. team training, documentation, workflow integration]	[DELIVERABLE — e.g. All users trained and active]	[e.g. Pilot sign-off]
Review & Optimise	[WEEKS]	[ACTIVITIES — e.g. measure results, gather feedback, optimise prompts]	[DELIVERABLE — e.g. ROI report and recommendations]	[e.g. 4 weeks of usage data]

Section 5: Risks & Mitigations

Risk	Likelihood	Impact	Mitigation Strategy	Owner
AI outputs contain inaccuracies	High	Medium	Mandatory human review before any output is used externally. Verification checklist for critical data.	[NAME]
Team resistance to new tools	Medium	High	Involve team in pilot design. Demonstrate time savings. Address concerns openly. Provide ongoing support.	[NAME]
Data privacy breach	Low	Critical	Enterprise-grade tools only. Data classification training. No PII in prompts. DPA with vendor.	[NAME]
Scope creep	High	Medium	Clear out-of-scope section. Change request process for new requirements. Regular scope reviews.	[NAME]
Tool vendor changes pricing or features	Medium	Low	Avoid vendor lock-in. Document prompts and workflows so they're portable. Annual vendor review.	[NAME]

Approval checklist: Before submitting this brief, ensure you have: (1) Quantified the problem, (2) Defined measurable success criteria, (3) Identified data requirements and privacy considerations, (4) Estimated realistic timelines, and (5) Documented risks with mitigations. A thorough brief dramatically increases your chances of getting stakeholder buy-in.