Back-office decision model · Legal 01

The deal is ready.
The contract is not.

Northline Benefits Technology has $8.4M in annual recurring revenue waiting in legal. One general counsel and one contract manager are reading the same liability, IP, privacy, and termination language again and again.

Open the risk workbench Read the case

// fictional company · illustrative data · no upload · runs in your browser

01 · Detect

Compare every clause with a visible company playbook.

02 · Decide

Separate safe redlines from issues that need human counsel.

03 · Evidence

Keep the source clause, rationale, change, and owner together.

Use-case story

One lawyer became
the company’s queue.

Northline is a fictional 720-person U.S. benefits-technology company. Growth brought larger enterprise buyers—and their paper. Legal review did not become harder because every agreement was unique. It became harder because familiar exceptions were buried in unfamiliar documents.

01 / MONDAY

Twenty-three agreements arrive.

Customer MSAs, data-processing addenda, vendor SLAs, NDAs, and partner terms all enter one shared inbox. Sales marks nine “urgent.” Procurement marks four more.

02 / WEDNESDAY

Simple and dangerous look alike.

An NDA needs one definition changed. A $1.8M customer MSA shifts unlimited consequential damages onto Northline. Both wait in the same queue for the same reviewer.

03 / FRIDAY

Counsel buys time, not a system.

Outside review absorbs overflow at an illustrative $420 per agreement. The invoice grows, but the approved positions and negotiation history still live across email and documents.

04 / WITH THE AGENT

Exceptions become a routed decision.

The AI extracts clauses, tests them against policy NB-LGL-2026.3, drafts approved fallback language, and sends only material exceptions to counsel—with an evidence trail attached.

Redline & risk matrix workbench

Change the policy. Watch the queue move.

Choose a contract, select a highlighted clause, and test three risk postures. Every number below is calculated from the fictional assumptions shown on this page.

Median review time

2.8 hr from 7.4 business days

Outside counsel

$9,568 per month · down $29,072

Auto-cleared

77 of 92 agreements / month

Legal capacity returned

340 hr per month · review labor model

Portfolio risk routed

7 + 6 counsel + redline flags in sample

Master Services Agreement

Counterparty draft · Fictional sample · Click a highlighted clause

What the number means

No savings number without its denominator.

This is an operating model, not a promise. The simulator separates volume, routing rate, review time, and outside-counsel cost so a buyer can replace every fictional input with observed company data.

Boundary: The agent supports issue spotting, policy comparison, drafting, and routing. It does not make final legal judgments, sign agreements, or replace licensed counsel.

Model input	Why it matters	Base value
Monthly agreements	Customer, vendor, privacy, NDA, and partner paper entering legal.	92
Current outside review	Illustrative blended cost applied to each agreement.	$420 / agreement
AI-routed counsel review	Share of agreements still requiring licensed counsel under the selected posture.	16%
Outside cost after routing	Referred agreements × illustrative $650 focused-review cost.	$9,568 / month
Internal review labor	Current 4.6 hours versus AI-assisted 0.9 hours per agreement.	340 hr returned
Platform cost	Excluded because product, implementation, integration, and security-review prices are unknown.	Not modeled

The operating change

Legal reviews the exception.
The system remembers the rule.

The first pilot should be smaller than the promise: one agreement type, one approved playbook, historical contracts with known outcomes, and a mandatory human sign-off. Measure precision, missed material issues, reviewer time, and negotiation acceptance before expanding scope.