---
title: McKinsey’s 2025 State of AI and the Case for Workflow Redesign
description: McKinsey’s State of AI emphasizes why support leaders must redesign workflows - and how PixieBrix’s Copilot turns AI into faster, complete resolutions.
---

[PixieBrix Blog](https://blog.pixiebrix.com/blog)

# [McKinsey’s 2025 State of AI and the Case for Workflow Redesign](https://blog.pixiebrix.com/blog/ai-support-copilot-mckinsey-2025)

 Written by [Eric Bodnar](https://blog.pixiebrix.com/blog/author/eric-bodnar) | Sep 9, 2025, 8:38:01 PM

McKinsey’s 2025 State of AI survey has one message support leaders can’t ignore: AI only creates value when workflows change. More than three-quarters of organizations now use AI in at least one business function, and genAI adoption has surged to 71%. Yet over 80% say they haven’t seen tangible enterprise-level EBIT gains. Why? Because tools alone don’t transform support - work design does.

## Why Support Leaders Struggle to See ROI

- **Governance gaps** → McKinsey shows companies with CEO-level AI governance see higher bottom-line impact. In support, this translates to clear ownership of data access, escalation policies, and risk controls.
- **Old workflows + new tools = friction** → Only 21% of orgs using genAI have actually redesigned workflows. Most teams just bolt copilots onto legacy triage and escalation steps. The result? More clicks, not fewer.
- **Value is local, not enterprise-wide (yet)** → Within support functions, AI already drives measurable ROI - faster handle times, fewer escalations, higher agent throughput. But those results don’t hit the CFO’s dashboard until they scale across the org.

## A Practical 4-Sprint Plan for Support Teams

**Sprint 1 - Baseline reality (2 weeks)**  
Map the true path of a support ticket across tools (Zendesk, Jira, Gmail, Slack). Measure time-to-first-useful-answer and escalation rates.

**Sprint 2 - Put AI in the flow (2–4 weeks)**  
Drop copilots directly in the sidebar where agents work. Provide contextual answers with citations, quick action buttons (update Jira, send Slack), and human-in-the-loop controls.

**Sprint 3 - Redesign workflows (4 weeks)**  
Eliminate one handoff. Automate environment collection. Draft replies and wrap-ups. Shift focus from “ticket deflection” to resolution quality and escalation reduction.

**Sprint 4 - Productize governance (ongoing)**  
Assign an exec owner. Set data access rules, monitoring, and rollback plans. Measure against three core KPIs: AHT, first-touch resolution, and escalation avoidance.

#### Key Stats to Ground the Conversation

- 71% of orgs now use generative AI (up from 65% earlier in 2024).
- 21% have fundamentally redesigned workflows.
- \>80% have yet to see enterprise-level EBIT impact.
- GenAI impact is strongest in service ops and software engineering - the very adjacencies of technical support.

#### The Support Leader’s Takeaway

The message for support leaders is clear: don’t just add AI, rewire support around it. Governance and workflow redesign are the levers that turn AI pilots into measurable outcomes. Start local - service ops is already where McKinsey sees impact - and scale out from there.

Ready to see this in action? Try the PixieBrix Customer Support Copilot - built for real support teams, not ticket deflection.

✅ DOM-aware context across Zendesk, Jira, Gmail, Slack  
✅ One-click actions to reduce escalations  
✅ Your choice of LLMs (bring your own OpenAI, Anthropic, or others)  
✅ Free through 2025 for Seed–Series B SaaS

[Apply for early access today.](https://www.pixiebrix.com/use-case/customer-support-copilot)

[View full post](https://blog.pixiebrix.com/blog/ai-support-copilot-mckinsey-2025)

```json
{
  "@context" : "http://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Eric Bodnar"
  },
  "dateModified" : "2025-09-09T20:38:05.610Z",
  "datePublished" : "2025-09-09T20:38:01Z",
  "headline" : "Rewiring Support for AI Value: Lessons from McKinsey’s State of AI",
  "image" : {
    "@type" : "ImageObject",
    "height" : 900,
    "url" : "https://47598997.fs1.hubspotusercontent-na1.net/hubfs/47598997/mckinseyblog.png",
    "width" : 1600
  },
  "mainEntityOfPage" : "https://blog.pixiebrix.com/blog/ai-support-copilot-mckinsey-2025",
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject",
      "height" : 60,
      "url" : "/hs/hsstatic/content_shared_assets/static-1.4092/img/default-amp-logo.png",
      "width" : 60
    },
    "name" : "PixieBrix Blog"
  }
}
```