What is the return on investment (ROI)

of generative AI in business?

Poorly framing your AI project

The first challenge that can cause the ROI of a generative AI initiative to melt away like snow in the sun is poor project scoping.

Before deploying AI in your organization, it is crucial to clearly identify the business needs this technology should address. For generative AI tools to be used by employees, they must solve a specific problem encountered by them. Without this, teams will not use the tools.

As an IT Manager, you must understand the expectations of employees. For example, if HR staff spend long hours answering employee questions about your company's compensation policy, implementing a chatbot that automatically handles employee inquiries could be a relevant solution.

To identify business needs and employee use cases, we encourage you to:

• Survey the relevant employees via an internal poll;
• Organize interviews and collaborative workshops with business teams;
• Map out existing processes to identify opportunities for improvement.

Poorly structured database

Another risk threatening the ROI of your generative AI projects is a poorly structured data set before AI integration. Generative AI that accesses incomplete or disorganized data is likely to provide inaccurate or misleading responses.

To ensure the effectiveness of your AI solutions, you should establish a data collection and management strategy in line with your company’s security and confidentiality policies.

Untrained employees

The third mistake that can reduce the ROI of your generative AI projects is the lack of internal user skills.

An employee who writes imprecise prompts, for example, may obtain superficial results that do not allow them to improve productivity. Worse, an employee who fails to verify the answers provided by the generative AI may overlook instances of AI hallucination, potentially leading to errors that could harm the company.

Imagine a network analyst in the telecommunications sector. If they request recommendations to optimize traffic distribution on a mobile network but formulate a vague prompt without targeting a specific issue, they risk receiving irrelevant or even counterproductive suggestions. Without thorough verification, these recommendations could lead to incorrect adjustments, degrading service quality and causing an increase in customer complaints, with detrimental consequences for the employer.

To prevent the lack of employee skills from negatively impacting the ROI of your AI projects, you can implement various training initiatives:

• Structured online or in-person training programs tailored to the employees' knowledge levels;
• A mentorship program between internal experts and employees new to artificial intelligence;
• Practical workshops where employees can work together on real use cases while being guided;
• A learning community where employees can share knowledge about AI (questions, advice, news, etc.) and maintain a continuous learning dynamic;
• Seminars and conferences led by experts to raise employee awareness of the challenges, opportunities, and innovations related to AI.

Ultimately, this learning journey should enable staff to craft effective prompts and challenge AI to mitigate the risks of AI hallucinations. But that’s not all. These initiatives should also reduce resistance to change and help teams understand the opportunities and risks (data security, ethics, etc.) associated with AI.

Streamlining IT ticket management with a chatbot

Generative AI improves the handling of IT support requests, a use case that can deliver good ROI. For example, if your organization uses a Microsoft-based intranet, you can implement an AI-powered, pre-prompted chatbot to instantly respond to employees’ IT issues.

In other words, to streamline IT ticket management in your company, you can deploy an internal AI-powered chatbot that will automatically respond to employees by leveraging your organization’s technical documentation. The result, according to a study conducted by Mozzaik with its clients, is a 50% reduction in the tickets handled by the IT team.

The outcome? Your IT team can focus its efforts on resolving more complex technical issues. Even better, with the faster resolution of system failures and bugs, all employees across the organization become more productive, and the overall employee experience is significantly improved.

Improving decision-making through data analysis

Improving decision-making processes is also a use case for generative AI that can provide a strong return on investment.

Generative AI solutions can leverage the company’s internal database and external data to anticipate market trends, forecast customer needs, and formulate strategic recommendations with a reduced error rate. In doing so, AI enables decision-makers and managers to make more informed decisions, seize opportunities quickly, anticipate risks, and ultimately gain a competitive advantage over their competitors.

In the industrial sector, for example, artificial intelligence can enhance decisions related to supply chain optimization. By analyzing real-time data on production, sales, suppliers, and the market, generative AI can estimate future demand, optimize order volumes and delivery times, and even anticipate issues in the logistics network before they arise. By basing their decisions on AI-generated predictions, leaders and managers can thus manage more accurately, improve operational efficiency, and reduce costs.

Facilitating document searches with RAG

According to a study conducted by ABBYY, 57% of French employees struggle to access the information they need to work when it is contained in documents. Additionally, 95% of employees report losing up to 8 hours per week searching for information. Unsurprisingly, using generative artificial intelligence to improve document search offers significant profitability.

This use case for generative AI in companies is based on RAG, or Retrieval-Augmented Generation. This technique combines two AI approaches: information retrieval from a database and text generation. In practical terms, RAG allows a language model like ChatGPT to search for information in documents and then generate a natural language response based on the retrieved information. The major advantage of RAG is that it delivers precise and reliable results. Why? Because it relies on specific sources rather than solely on data generated from the model’s training.

For companies, RAG represents a small revolution. It enables employees to spend less time searching for information within their organization’s documentation (reports, archives, technical documentation, etc.). With RAG, employees can "query" a document using natural language and receive responses that are also formulated in everyday language. No more reading through documents one by one—just ask their research assistant a question, and they’ll instantly access the relevant information.

Does this sound a bit abstract? Imagine a consultant who needs to quickly verify some information to finalize a client presentation but isn’t sure exactly where to find it in the company’s documentation. This employee can use Genius by Mozzaik, their work assistant integrated into their Digital Workplace and powered by ChatGPT-4, to obtain the desired information instantly. All they need to do is ask their chatbot. The chatbot will then provide a precise and reliable response based on internal sources. And if our professional wants to review a lengthy study of several dozen pages just before a meeting, that’s also possible. RAG technology can summarize a document or extract its key themes in just a few moments. The result? A significant productivity boost and an improved employee experience.

Accelerating document creation

Content production is one of the use cases where generative AI offers the best return on investment for companies.

With generative artificial intelligence, employees can accelerate their creative process and even produce better content. Solutions like Genius by Mozzaik, for example, help them find new sources of inspiration by suggesting fresh ideas based on their previous projects. Genius by Mozzaik enables employees to challenge their ideas, pushing them to continually innovate and exceed expectations.

This use case applies to all teams that create documents such as reports, presentations, creative briefs, training programs, articles, or social media posts. The marketing department, for instance, can leverage AI to reduce the time spent on producing product descriptions for the company's website. They can ask the generative AI to analyze images of these products and generate relevant metadata.