The State of AI in Business 2025

A recent MIT report suggests that 95% of organizations are getting zero return from their generative AI investments.

The headline figure might suggest that you should be pessimistic or hesitant about your own journey into the AI future.

Yet the devil is in the detail. The decision should not be about whether you invest in AI initiatives, but how.

Successful companies are focused on buying market-leading platforms (such as ChatGPT) and less on building or customizing their own internal tools, which are often rejected by employees.

Prefer to watch?

Why it matters

Every single company and every single industry is undergoing rapid change as AI disrupts every process, every role, and every customer experience.

Positioning your company on the right side of this change today will determine your ability to compete tomorrow.

Not since Y2K and the millennium bug has every company had to do exactly the same thing by exactly the same deadline or risk extinction.

Companies that hold on to the idea that AI is a technology project, an IT initiative, and double down on building their own in-house tools for employees, will lose time against those that accelerate using the tools that are already in their employees' hands.

Employees are in charge (of what they use)

In 2005, if your company made the decision to roll out Microsoft Dynamics for your CRM, there was no possibility of you going home and spinning up your own instance of Salesforce.com to manage your pipeline. You had to use what the company provided.

But in 2025, that is not the case. If your company mandates the rollout of Microsoft Copilot for security and licensing reasons, you still have ChatGPT, Claude, or Grok on your personal phone, on your personal laptop, and you will use the tool that you think will help you get the job done best.

Let me give you two charts from the report to bring this to light.

This chart shows that only 40% of companies have purchased a corporate LLM subscription. We can think of Copilot, Gemini, ChatGPT Team and ChatGPT Enterprise within this group.

Yet 90% of employees say that they're using LLMs regularly - this suggests that a significant proportion of employees are relying on personal AI tools, often outside of formal company provision.

In this second chart, users were asked why they preferred to use a generic tool like ChatGPT vs an internal enterprise tool that had been built for a specific purpose.

State of AI in business - user preference drivers.

The responses show employees are choosing to use the familiar tool (ChatGPT) that's already on their phone and that they're using for multiple other use cases, both at work and in their personal lives.

The report mentions one example of a corporate lawyer at a mid-sized firm whose organization had invested $50,000 in a specialized contract analysis tool. Yet she consistently defaulted to ChatGPT for drafting work:

"Our purchased AI tool provided rigid summaries with limited customization options. With ChatGPT, I can guide the conversation and iterate until I get exactly what I need. The fundamental quality difference is noticeable, ChatGPT consistently produces better outputs, even though our vendor claims to use the same underlying technology."

I'm often asked by Kowalah clients, why it is that we recommend rollouts of ChatGPT Team and Enterprise over a deployment of a specialized tool, or the default Copilot, or Gemini linked to their enterprise Microsoft or Google agreement.

The reality is that a $30 licence for ChatGPT, complemented by its Deep Research, its GPTs, its voice, its projects, its images, its memory, its mobile apps, often outperforms any tool that you may have invested tens or hundreds of thousands of dollars in.

And employees will pick up the tool that is best for the job.

Start lighting a thousand little fires

Our mantra to clients is: instead of trying to light one big fire (i.e., a centralized enterprise 'build your own' AI project),

we recommend lighting 'a thousand little fires' where you give every single one of your people the skills and confidence to embed ChatGPT and AI thinking into every little micro-piece of work that they're doing.

Firstly, make a quick decision about which vendor you believe provides the best functionality and applications built on top of their model today and has the best capability to develop that roadmap over the coming two years.

Secondly give that platform to every single one of your people and train them on how to embed that tool into their day-to-day work:

planning for their week
preparing for a meeting
working with colleagues
hiring a new recruit
building their annual plan

Thirdly, as an IT team, focus on supporting these thousand little fires with pro-user capabilities such as:

Developing GPTs to share commonly used prompts across the organization
Creating projects to retain memory around common tasks and processes
Building custom actions in GPTs to allow them to have access to and to update other systems
Building mini web applications that can allow external parties (including prospects and customers) to interact with your AI workflows

Instead of trying to fight the free tools that your employees are choosing to use, embrace them, bring them under your security and governance with a corporate account.

Go deeper

There are other great insights in this 26-page report, and I encourage you to read it.

Download Here

Just be sure to go further than the headline that 95% of Gen-A projects are failing.

How can Kowalah help?

No doubt after reading reports like this you have questions about the implications for your own organization.

Kowalah is your digital Chief AI Officer here to help you think through your AI strategy, to pick and implement the right AI tools, and to develop and deploy your organizational change program.