Internal · Pre-kickoff · PM Series

Read the SOW
before it reads you.

Every project arrives with a signed contract, approved hours, and a timeline. The SOW Audit reads it from three simultaneous perspectives — Lead Designer, Project Manager, and Program Manager — and tells you exactly what will hurt before the kickoff call.

Sales closes. CEO approves for margin. Delivery inherits the gap. This tool exists because that gap has a name — and it shows up in every project, always in the same places.

3
Perspectives
simultaneously
4
Output sections
per audit
0
Diplomacy.
Brutal is the point.
01 Why this exists

The same four problems.
Every single project.

These aren't hypothetical risks. They're documented, named, and repeating. The SOW Audit was built directly from evidence across real NetBramha projects — each one a version of the same failure mode.

Corteva
13 gaps flagged after sign-off
Wrong project references, no page cap, vague deliverables, copy-paste errors from other engagements. All caught manually after the SOW was signed.
Godai
10 gaps raised internally, acknowledged, not fixed
PM flagged missing approval process, no iteration limits, no scope freeze clause. Sales responded "will incorporate going forward." Nothing changed.
PrismForce
Screen cap hit and exceeded
25 unique screen cap in the SOW. Team delivered more than agreed. Client kept requesting additional journeys. No contractual protection. The conversation dragged for months.
Websol
6-week project ran to September
Fixed timeline, client feedback arriving a month after contracted end. No client delay clause. No auto-approval mechanism. "Recommended 48 hours" is not enforceable.
02 How it works

Three seats.
One honest read.

The same SOW looks different depending on where you're sitting. The Lead Designer sees a screen count problem. The PM sees an unenforced approval window. The Program Manager sees hours that can't deliver the scope. The audit reads it from all three simultaneously.

Lead Designer
Hari's seat
Screen count, persona complexity, iteration limits, illustration scope, design system depth, prototype fidelity. Catches "6 personas across mobile and web" before anyone else does.
Project Manager
Sharath's seat
Approval mechanisms, feedback windows, change request clauses, screen caps, iteration limits, client accountability. Knows "recommended 48 hours" is not a contract term.
Program Manager
Sarath's seat
Commercial reality. Whether approved hours can deliver the stated scope. Whether the billing model matches the risk profile. Whether the SOW protects NetBramha when things go wrong.
↓ What goes in
📄
Design SOW
PDF · Required
📋
Delivery SOW
PDF · Optional
⏱
Approved Hours
Design · Delivery · Dev
💬
Context
Verbal commitments, risks

The approved hours are the most important input alongside the SOW — because that's where the formula disconnect lives. Claude needs to know what sales committed to before reading what the SOW actually contains.

↓ What comes out
01
What the SOW actually says
In scope, out of scope, vague zones, missing clauses, assumption landmines, and cross-SOW conflicts. Everything surfaced — not just what's written, but what's missing and what's contradicted between documents.
Extraction
02
What will hurt — and when
Every risk flagged with severity (Critical / Amber / Watch), category, which seat it comes from, what actually happens in delivery, when it surfaces, and the exact clause that creates it.
Risk Map
03
Questions to raise in kickoff
Not open-ended questions — specific alignment points to close before work begins, each with what to get in writing. The PM decides what to raise and when.
Kickoff Prep
04
For Sales / CEO
A tight paragraph written for a non-delivery audience. Frames gaps as delivery readiness concerns, not blame. Professional tone. PM edits before sharing — or doesn't share at all.
Shareable
03 Who uses it

Same output.
Different conversations.

The tool produces one honest view. What happens with it depends on the room. PMs run it internally and decide what to surface, when, and to whom.

Program Manager
Runs the audit
Gets the full brutal output before the kickoff call. Knows exactly which gaps to close, which questions to raise, and what to get in writing before the team starts work.
Lead Designer
Design-seat perspective
Can run the audit independently or review the output. The Designer seat catches effort mismatches that the PM might not flag — screen complexity, iteration depth, system scope.
CEO / Sales
Receives the summary
Gets the shareable paragraph at the bottom — delivery readiness framing, not blame. The PM decides what to show and when. One button to copy to clipboard.
Delivery Team
Inherits the brief
The kickoff prep section is designed to be actionable for the team walking into the first meeting. Specific questions, specific confirmations, what to get in writing before day one.
04 The stack

One new skill.
Everything else familiar.

SOW Audit uses the same stack as the rest of the PM series — with one new capability: sending PDF documents directly to Claude as base64-encoded blocks. No server-side parsing. No text extraction step. Claude reads the SOW natively.

🧠
Anthropic Claude
Reasoning engine
Reads both SOW PDFs simultaneously, audits from three seats, returns structured JSON with extraction, risks, kickoff prep, and verdict.
📑
Anthropic PDF API
Native document reading
PDFs are base64-encoded in the browser and sent directly in the API call. Claude reads layout, tables, and structure without any separate extraction step.
🗄️
Supabase
Audit history
Every audit stored in sow_sessions and sow_outputs tables. Shared project with RtR, SHI, and Career Bridge — cross-product analytics possible.
🌐
Netlify
Hosting
Single HTML file. Drag-and-drop deploy. Live in under 2 minutes. Same approach as every other tool in the series.
📊
Supabase History
Pattern detection
Because every audit is stored, future queries can identify which risk categories surface most often across all projects — turning individual audits into organisational learning.
🔮
What's next
Scope Change Evaluator
Natural extension — evaluate a client change request against the original SOW. Same PDF input pattern, same three-seat analysis, different prompt.
SOW Audit · PM Series 08 · NetBramha Studios
Sarath MS · Program Manager