How Generative AI Is Changing Business Workflows

McKinsey estimated that usage of generative AI in at least one business function increased from 55% to 78% of businesses over a span of one year. It's a really quick uptake. But how generative AI is changing business workflows is a question that differs from how many companies are using it  and the statistics don't look so good for the latter: McKinsey's own research found about the same percentage of companies are seeing "material impact" in the bottom line.

This guide will explain what verified data actually means: where generative AI is delivering workflow improvements and why most companies aren't seeing them and how the two differ.

Understanding The Concept Of Generative AI In Business

Generative AI in Business is AI systems that generate content, code or analysis based on a prompt, rather than those previous AI business automation systems which followed logic based on rules. The current wave of AI workflow automation, across most industries, is centered on the ability of Generative AI technology to take multiple steps within a specific workflow, such as drafting communications, writing and reviewing code, summarizing documents, and much more.

The difference is important because most of the generative AI business applications today are "horizontal," meaning that they're business tools like chatbots and writing assistants that are used in a variety of jobs. While these represent the most popular generative AI use cases, they are not the ones that have seen the highest dollar benefit, which is precisely why a more generative AI solutions around a specific task will always be more beneficial than a generic deployment of generative AI.

The Gen AI Paradox: Wide Adoption, Uneven Impact

This is what McKinsey, itself, calls the “gen AI paradox”: almost 8 out of 10 companies have used some version of generative AI, but about as many say the technology has had a “significant impact” on their bottom line. Technology was adopted almost universally before it was of measurable value.

According to an article published in the Harvard Business Review in September 2025, that ratio was even more dramatic: 95% of companies report no quantifiable benefit from their use of generative AI. Two independent research groups, with different methods, came to the same conclusion: Adoption and impact do not equal one another. That's why it's important that people arriving at the same result from independent studies of different company samples reach the same conclusion.

Verified Data Behind Generative AI Business Applications

The singular explanation provided by BetterUp Labs and Stanford is the useful finding "workshop" – the more AI uses to generate content with little or no added value, the more real and measurable downstream cost it has   in this case, nearly two hours of rework per occurrence. When rework is applied across an entire team on a daily basis without a review step and multiplied by the time that the generative AI tool saved, the rework time adds up fast enough to offset the initial time saved.

Where Generative AI Automation Is Currently Being Applied

Whether a business is exploring generative AI for business use in engineering, operations, or customer-facing functions, the productivity improvements that are measurable often stem from the application of generative AI automation to specific, well-defined technical tasks.

1. Software Development

According to a study by MIT Sloan, Microsoft Research and GitHub, generative AI coding tools reduced the time required for coding by 56%, one of the most measurable productivity benefits of generative AI technology in any task that was recorded, reported in Harvard Business Review.

2. Written Communication & Drafting

The same 2023 study showed that generative AI can help individuals complete an internal analysis or any kind of email 40% faster than they were before, which is an obvious step-by-step improvement by generative AI for business that any business can easily see

3. Work Activity Augmentation at Scale

According to Accenture Research, cited in the Harvard Business Review, over 40% of all U.S. work activity can be enhanced, automated, or reimagined using generative AI, with the most significant expected impact in legal, banking & insurance, and capital markets sectors; followed by retail, travel, health, and energy.

How To Differentiate Between AI Workflow Optimization And Ai Workflow Sprawl?

78% adoption versus 95% no-ROI is not a technology problem, it is a workflow design problem. Like most AI-powered workflows, effective generative AI solutions were developed from a rethought workflow, rather than added on to an existing one.

• The goal of AI workflow automation is to address a specific measurable task, not a generic directive to use AI more  such as coding or writing the examples above.
• It also has a review or verification step, which helps to identify and address the "workslop" pattern that generates additional rework time than it saves.
• It calculates a real business impact time saved, defect rate, cost per task etc., rather than how many times the tool is used.

Benefits Of Generative AI For Redesigned Workflow

The following are a few benefits that generative AI provides in a reshaped workflow.

As the conditions outlined above are met, the benefits of generative AI are real and specific, as cited throughout this guide: Coding time cut by 56%, Writing speed improved by 40%, and Generative AI can augment 40% of all U.S. work activity. In this case, it was not a workforce that was being replaced, it was a specific task that was being redesigned around the strengths of well-designed AI-powered workflows, as is the case in every example of generative AI productivity that we've explored in this guide.

Where AI/ML Development Services Fit Into Workflow Redesign Process

For AI/ML development to become impactful and measurable for a business, the gap between adopting AI/ML and achieving the impact typically needs to be bridged with some workflow redesign, integration into systems, and review steps to avoid the 'workshop' pattern covered in this guide.

Companies considering a partner for this endeavor should seek out a provider that defines success after deployment specifically and not merely the result of a demonstration test, which is a difference between those that McKinsey rates as high performers and the 95% who are reporting no measurable return. To anybody who cannot explain what would happen to the workflow 6 months after its launch, it is a pilot, not a production system.

How Chirpn Supports Generative Ai Workflow Projects.

Chirpn IT Solutions develops AI-powered workflows, and even more, entire AI business automation systems, that are centered on a particular, measurable business outcome and are not just a generic AI feature, which is being added to an existing process. Chirpn's involvement with Talent 100, an Australian education company, has involved AI/ML development as part of a platform build, which has been completed on the standard 45-60 day delivery timeline – such work being handled using Chirpn's AI orchestrated platform development framework, AutoPATH.

In fact, the most important success factor of a generative AI tool isn't AutoPATH-driven development, it's Chirpn's API and systems integration work, which puts the generative AI tool in the hands of teams that use the system every day. As a certified Google Cloud Partner, equipped with the tools and services of Google's Vertex AI and Google Agent Assist, Chirpn adds production-grade AI infrastructure to workflow automation projects that are tied to a specific, measurable outcome.

Conclusion

How generative AI is changing business workflows depends entirely if the new tool can be integrated into a current workflow, or can act as a catalyst to rethink an entire process? The verified data aligns with several independent sources: In a year, adoption has increased from 55% to 78% of companies, but an equal percentage say no financial impact, and research at MIT Media Lab estimates that number could be as high as 95% of companies.

The companies that are filling that gap are the ones using AI in specific, well-delineated tasks where there's a review step and there's a clear outcome  not the ones using AI the most, but the ones using it for specific, well-delineated activities, where they have a measurable outcome and a review step. But that's a distinction, rather than a capability of a single tool, that will make or break if generative AI is truly going to transform a business's processes or simply another buried functionality. The research quoted in this guide has come from many different angles to the same conclusion and that is why it is worth considering as a genuine planning input and not just another data point to avoid.

Frequently Asked Questions

How is generative AI changing business workflows?

According to research cited in Harvard Business Review, generative AI is having the greatest impact on business workflows when it comes to software development, where coding tools cut programming time by 56%, and in written communication, where tools can help people draft content approximately 40% quicker. More widespread transformation of workflows hasn't been uniform. McKinsey reported that 78% of companies had adopted it but did not report any significant effect on bottom-line results.

Why do most businesses fail to get ROI from generative AI?

The answer lies in a key factor: 41% of workers reported low quality and unchecked AI output that needed almost two hours of rework on average, according to A. MIT media lab research, which revealed that 95% of organizations report no measurable ROI from the use of generative AI. In fact, Harvard Business Review's coverage of BetterUp Labs and Stanford research revealed that this is the problem. The most common cause for lack of ROI is not going through the workflow redesign and review process.

What are the biggest generative AI use cases in business today?

The generative AI use cases that are most measurably successful are narrow, technical and focused on software development (56% faster coding), written communication and drafting (40% faster), and broad work-activity augmentation – in which more than 40% of all U.S. work activity was found to be augmented or automated by generative AI, according to Accenture Research.

What is AI workflow optimization, and how is it different from just using AI tools?

AI workflow optimization is not simply about integrating an AI tool into a business process, but about transforming the business process around the capabilities of AI. There is a measurable difference: The research conducted by McKinsey reveals that the distinction between companies that use generative AI (78% of them) and those that report a real financial impact is this one.

Who is an ideal partner for a business when it comes to generative AI automation?

Don't just ask how a demo does, ask how a potential partner does. Regardless of how well they've crafted their initial pitch or demo, a partner who can't tell you what they're going to get back from generative AI has likely not solved this issue yet, based on the MIT Media Lab's research that finds 95% of organizations report no measurable ROI from generative AI.