---
title: AI can make the wrong work happen faster · Zero Fog
description: "AI changes the economics of work but has no view on whether the work should exist. Why order matters: see it, remove, fix the joins, then accelerate."
image: https://www.zero-fog.com/hubfs/perspectives/zf-lesson-04-featured.png
---

[![Zero Fog](https://www.zero-fog.com/hubfs/Zero-fog-site/zero-fog-logo.png)](https://www.zero-fog.com/)

[How we work](https://www.zero-fog.com/how-we-work) [Situations](https://www.zero-fog.com/situations) [Selected work](https://www.zero-fog.com/reference-cases) [Perspectives](https://www.zero-fog.com/insights) [About](https://www.zero-fog.com/about) [Talk to us](https://www.zero-fog.com/contact)

[Perspectives](https://www.zero-fog.com/zero-fog)

# AI can make the wrong work happen faster

Jayven Rappa · November 11, 2026

AI is very good at making work faster. It has no opinion about whether the work should exist.

AI genuinely changes the economics of work. Tasks that took days take minutes. Analysis that needed a team can be drafted by one person. The question is where it gets pointed. Too often it's aimed at the work as it runs today, with all the rework, waiting, duplicate checks and workarounds still in it.

## Faster isn't the same as better

Most workflows carry steps that exist only because of an older constraint: a system that couldn't hold a piece of data, a control added after one bad incident, a handoff designed around a team that has since moved. Accelerate that workflow as it is and every one of those steps gets faster too. Output goes up. So does the volume of rework. Errors arrive sooner and at greater scale.

From a distance this looks like success, because activity rises. Up close, the binding constraint hasn't moved, and it often wasn't in the part the technology touched.

![Two ways to apply AI to an eight-step workflow. Accelerating it as it is speeds up the duplicate checks, rework and waiting too. Deciding the work first removes three steps and accelerates the five that remain.](https://www.zero-fog.com/hs-fs/hubfs/perspectives/zf-lesson-04-diagram.png?width=760&height=309&name=zf-lesson-04-diagram.png)

*Speeding up the work as it is scales the waste along with it.*

## Where the constraint actually sat

A large bank set out to move tens of millions of clients from static product marketing to one-to-one personalization. The obvious move was to buy a platform and roll it out everywhere. Mapping the work showed why that would stall: data, content and decision logic sat in three separate teams, so no single client journey could be personalized end to end. The constraint was how those teams were joined, not the technology. Going journey by journey, highest value first, let each wave prove itself against measured sales before the next was funded.

A global strategy firm wanted its senior people to produce partner-grade analysis faster. Generic AI was quick, and partners didn't trust it. The fix wasn't a better model. It was a structured picture of the companies for the model to reason against, and output that said where it was confident and where it wasn't. A cycle that had taken about three months came down to about a minute and a half, and the depth held up.

## What the AI needs around it

When AI disappoints in a business, the model is rarely the weak link. What's thin is everything around it. I think of four things it needs: what *constrains* it (the rules and standards the work must meet), what *focuses* it (the decision or workflow it serves), what it *experiences* (a structured picture of how the business actually runs), and what it *learns from* (feedback tied to outcomes the business already measures). Most organizations invest heavily in the tool and lightly in those four.

## The discipline

The discipline is mostly about order:

1. **See how the work really runs**, across process, data, systems, people and culture.
2. **Remove what shouldn't exist** before making anything faster.
3. **Fix the joins**, the handoffs and decisions between teams, where the constraint usually sits.
4. **Then accelerate what remains**, and measure it against an outcome the business already tracks.

Done in that order, the AI investment is usually smaller, faster to prove and easier to defend.

## A test for your next AI decision

Before approving it, try to answer four things in a sentence each: which workflow it changes, where that workflow stalls today, which measured outcome should move, and the smallest first step that would prove it. If those are hard to answer, the decision isn't ready yet, however good the technology is.

*Lessons from the Fog is a series on where capable organizations lose sight of how they really work, and the disciplines that bring it back into view. Previous lesson: [A dashboard can tell you what's happening without telling you why](https://www.zero-fog.com/perspectives/dashboard-what-not-why).*

**Facing a consequential decision?**

A 30-minute conversation about the decision ahead, whether it fits, and what a first cycle would cover.

[Talk to us](https://www.zero-fog.com/contact)

## Read on

[![](https://www.zero-fog.com/hubfs/perspectives/zf-lesson-03-featured.png) Nov 4, 2026A dashboard can tell you what's happening without telling you why](https://www.zero-fog.com/zero-fog/dashboard-what-not-why) [![](https://www.zero-fog.com/hubfs/perspectives/zf-lesson-02-featured.png) Oct 28, 2026Your organization chart isn't your operating model](https://www.zero-fog.com/zero-fog/org-chart-isnt-your-operating-model) [![](https://www.zero-fog.com/hubfs/perspectives/zf-lesson-01-diagram.png) Oct 23, 2026The most expensive thing in business might be what you think you know](https://www.zero-fog.com/zero-fog/what-you-think-you-know)

[![Zero Fog](https://www.zero-fog.com/hubfs/Zero-fog-site/Transparent%20Logo_white%20logo%20side.png)](https://www.zero-fog.com/)

Founder-led advisory for consequential decisions.

Site

[How we work](https://www.zero-fog.com/how-we-work)  
[Situations](https://www.zero-fog.com/situations)  
[Selected work](https://www.zero-fog.com/reference-cases)  
[Perspectives](https://www.zero-fog.com/insights)  
[About](https://www.zero-fog.com/about)

More

[Partners](https://www.zero-fog.com/partners)  
[Rejoyce](https://www.rejoyce.ai)  
[Talk to us](https://www.zero-fog.com/contact)

Contact

hello@zero-fog.com

 © 2026 Zero Fog, LLC. All rights reserved. · Massachusetts, USA

[Privacy](https://www.zero-fog.com/privacy) · [Terms](https://www.zero-fog.com/terms) · [LinkedIn](https://www.linkedin.com/company/zero-fog)

```json
{
  "@context" : "https://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Jayven Rappa",
    "url" : "https://www.zero-fog.com/zero-fog/author/jayven-rappa"
  },
  "dateModified" : "2026-11-11T00:33:06.128Z",
  "datePublished" : "2026-11-11T00:32:08.000Z",
  "headline" : "AI can make the wrong work happen faster · Zero Fog",
  "image" : [ "https://www.zero-fog.com/hubfs/perspectives/zf-lesson-04-featured.png" ],
  "mainEntityOfPage" : {
    "@id" : "https://www.zero-fog.com/zero-fog/ai-wrong-work-faster",
    "@type" : "WebPage"
  },
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject"
    }
  }
}
```