---
title: "Stop bolting AI onto broken workflows"
description: "5 source-backed AI claims from Stop bolting AI onto broken workflows, with key statistics, context, and the original AI Adopters Club source."
url: "https://kbanc.com/claims-library/stop-bolting-ai-onto-broken-workflows"
source: "https://aiadopters.club/p/ai-on-human-workflows"
date: "2026-05-05"
topics: ["strategy", "implementation", "business"]
generated: "2026-08-31"
---

# Stop bolting AI onto broken workflows

By Kamil Banc | May 5, 2026

## Claims

1. **21% Workflow Rebuild Rate** (source summary): Only 21% of organizations have rebuilt their workflows around AI, according to McKinsey's 2025 research.
2. **Redesign Multiplies Revenue** (source summary): Startups trained in workflow reorganization generated 90% higher revenue with equal AI access and training.
3. **JPMorgan's Legal Hours Saved** (source summary): JPMorgan Chase's COiN platform saved 360,000 legal hours annually, returning $2 billion in 2024 benefits.
4. **PwC Productivity Gains** (source summary): PwC reported 20% to 50% productivity gains after resequencing workflows around human judgment.
5. **Failures Are Organizational** (source summary): Failures in AI initiatives are 70% organisational rather than technical, per the article.

## Evidence

### Quote
> "Tools are cheap, redesign is the moat." - Kamil Banc

### Key Statistics
- **21% of organizations have rebuilt workflows around AI**: McKinsey's 2025 workplace AI research; the remaining 79% bolt AI onto legacy processes, with half of initiatives stuck in pilot mode.
- **90% higher revenue**: INSEAD/Harvard randomized field experiment where 515 startups received equal AI access; only the workflow-reorganization training group achieved this uplift.
- **360,000 legal hours saved annually**: JPMorgan Chase's COiN platform; the firm's $2 billion AI spend returned $2 billion in benefits during 2024.
- **Three to five times the returns**: Firms investing over 20% of digital budgets in workflow redesign outperform firms that do not, per the article.

## Context
The article synthesizes findings from McKinsey's 2025 workplace AI research, an INSEAD/Harvard randomized field experiment with 515 startups, and a California Management Review framework called DGRI published in late 2025. The DGRI method prescribes five stages: diagnose and align, establish governance, redesign for scalability, reuse assets and build literacy, and iterate through minimum viable transformations. Practitioners are advised to select one workflow with a clear outcome metric, form a digital-steward team pairing domain and technical owners, and tie bonuses to workflow outcomes rather than tool usage. Case evidence from JPMorgan Chase, PwC, Mayo Clinic, and a global manufacturer illustrates that operating-model redesign, not tool procurement, drives measurable EBIT impact.

## Source
- Original: [Stop bolting AI onto broken workflows](https://aiadopters.club/p/ai-on-human-workflows)
- Cite: kbanc.com/claims-library/stop-bolting-ai-onto-broken-workflows

## Primary Evidence
- [That group generated 90% higher revenue, found 44% more use cases, won 18% more paying customers, and needed 40% less capital](https://www.uxtigers.com/post/workflow-redesign) (uxtigers.com; supports claim 2)