---
title: 'JPMorgan Chase: COiN agreement extraction'
slug: jpmorgan-coin-contract-review
stable_id: 8cd32f6eda7ef8fb
company: JPMorgan Chase
function_code: credit
pattern_codes:
  - exception_based_operations
  - intake_redesign
evidence_strength: reported
publication_tier: showcase
freshness: current
reviewed_at: '2026-08-23'
updated_at: '2026-08-22'
source_quality_summary: 1 independent, 1 primary; publication outcomes are reported.
caveat_summary: >-
  Company upper-bound estimate; seconds is machine extraction, not end-to-end
  disposition.
collections:
  - human-still-decides
  - queue-eliminated
  - regulated-autonomy
bundle_version: 1.0.0
bundle_fingerprint: sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93
canonical_url: https://brianletort.ai/transformations/jpmorgan-coin-contract-review
---

# JPMorgan Chase: COiN agreement extraction

Defined attributes can be extracted across annual volume while lawyers handle interpretation exceptions.

Function: Credit and lending. Patterns: Exception-based operations; Intake redesign. Evidence: reported.

Freshness: current. Reviewed: 2026-08-23. Updated: 2026-08-22.


Source quality: 1 independent, 1 primary; publication outcomes are reported.

## Before

1. **Lawyers and loan officers** — Read commercial-credit agreements and manually identify relevant clauses and attributes. (control: Approximately 12,000 annual agreements; company estimated up to 360,000 aggregate hours.)
2. **Loan-servicing staff** — Use interpreted terms in servicing processes and correct interpretation errors. (control: Human legal interpretation and servicing controls.)

## After

1. **COiN** — Uses unsupervised machine learning to extract 150 defined attributes from agreements. (control: Defined attribute schema; machine extraction reported in seconds, not end-to-end disposition.)
2. **Lawyers and loan officers** — Review exceptions and retain legal interpretation and servicing decisions. (control: Unknown or low-confidence clauses require human review; no public error-rate threshold.)

## Decision rights

Humans own legal interpretation and downstream servicing.

## Exception path

Unknown clauses or low-confidence extraction require human review.

## Outcomes

- **Processing effort** (reported): As many as 360,000 hours/year → 150 attributes from 12,000 agreements extracted in seconds. Company upper-bound estimate; seconds is machine extraction, not end-to-end disposition.

## Executive lesson

Constrain AI to a defined extraction schema.

## Anti-pattern

Do not imply complete lawyer replacement.

## Questions for leaders

- Where is the operating threshold set and who can override it?
- What measured result would trigger rollback or retraining?
- Which residual decisions must remain human-owned?

## Sources

- [JPMorgan Chase 2016 Annual Report](https://reports.jpmorganchase.com/investor-relations/2016/pdf/ar2016-lettertoshareholders.pdf) — JPMorgan Chase
- [JPMorgan Software Does in Seconds What Took Lawyers 360,000 Hours](https://www.bloomberg.com/news/articles/2017-02-28/jpmorgan-marshals-an-army-of-developers-to-automate-high-finance) — Bloomberg