Pourquoi le rapport du MIT sur l'IA n'est pas une surprise

A new report from MIT has made headlines, stating that 95% of company AI projects are failing to deliver real value (*). For companies that have invested millions in AI, this is shocking news. But for us at LumApps, it is not a surprise. It confirms what we have always believed about technology in the workplace: the employee experience is everything.

The problem is not the AI technology itself. The potential of AI is incredible. The problem is how companies are trying to use it. They are rushing to buy the latest AI tool, giving it to employees, and expecting magic to happen. But employees don't need another complicated tool to learn. They need their work to be simpler, and their daily experience to be better.

The Real Problem: A Bad Experience is a Failed Project

The MIT report talks about a "learning gap." We believe it is really an "experience gap." Most employees are busy. They do not have time to become experts in writing perfect "prompts" for an AI. They have a job to do, and they need help now. Expecting them to learn a new, complex skill just to use a new tool is a recipe for failure.

The report also found that many employees use their own AI tools secretly. This "Shadow AI" is a powerful signal. It shows that employees want to use AI to help them, but the tools their company provides are not good enough. This is a huge risk. When employees use unapproved tools, it can create serious problems for company data security and create inconsistent information across the company.

This chaos happens because there is no central strategy. Different departments buy different AI tools, creating a messy collection of bots that don't work together. This is why 95% of these projects fail to deliver any real, measurable business impact.

A Better Way: AI Inside Your Employee Hub

So, what is the solution? As we wrote in our blog post, "Will Agentic AI Kill the Intranet?", AI will not replace the intranet. It will make the intranet the most important technology system for a company. The employee hub is the perfect place to bring all your AI tools together in one simple, secure, and unified experience. It should be the "AI Front Door" for your business.

For AI to be successful in the workplace, it must follow a few simple rules.

‍

‍

1. The best AI is very specific.

A general AI that tries to do everything is not very useful for business. A specific AI that does one thing very well is much better. For example, at LumApps, we are building agents to help with specific company tasks. Imagine an AI that can instantly answer an employee's question about the company's vacation policy because it only uses the official, up-to-date HR documents. That is much more useful and trustworthy than a general chatbot that might get the answer wrong.

2. The best AI is ready to use and "behind a button."

Employees should not have to think about AI. It should just work, seamlessly integrated into their daily tasks. For our customers in internal communications, we are building tools to make their job easier. Instead of asking them to write a complex prompt, we give them a simple button that says "Refine Style" : with one click, the AI can suggest a better title for their article, create a short summary for a mobile alert, or even adapt the tone for different audiences. It's simple, fast, and helpful.

3. The employee must stay in control.

AI is a powerful assistant, not a replacement for human expertise. It can give suggestions, but the human expert should always make the final decision. When our AI helps create content, such as a weekly digest or a newsletter, it provides a draft. The communication professional then checks, edits, and approves it. This ensures the information is accurate, the company voice is right, and it builds trust in the technology.

Stop Buying Tools, Start Building an Experience

The message from the MIT report is clear: stop buying more separate AI tools. Instead,I focus on creating one simple, central AI experience for all your employees.

The future of work is not about everyone becoming an AI expert. It is about making work simpler by putting smart, specific, and easy-to-use AI inside the platforms employees use every day. This is how you move from failed experiments to real business transformation. And that is the future we are building at LumApps.

(*) You can read the full MIT report analysis in Fortune here: MIT report: 95% of generative AI pilots at companies are failing

‍