Launching an AI pilot is relatively straightforward. But integrating it reliably and at scale into business operations is a far greater challenge. McKinsey’s global 2026 survey found that only 37% of respondents said AI had improved their company’s earnings, unchanged from the previous year. Yet eight in ten said AI had helped them become more productive at work.
An AI tool may work well during a small test, when the task is simple and the data is clean. But using it across a company brings new challenges. It must integrate with existing software, comply with data access and security requirements, and have a clear process for handling failures. Someone must also take responsibility for its performance. Meeting these requirements means coordinating across IT, operations, finance, and risk teams.
Costs create another challenge. Companies are used to setting budgets for software projects and estimating staff time. AI costs can be harder to predict. They depend on how much information the system processes, which model it uses, and how often people use it. Without a clear estimate, finance teams may struggle to judge whether a project is worth the investment.
Cost each use case first
Zacharie Lahmi, founder of Ellis Partners, recommends that companies assess AI opportunities one business process at a time. He has ten years of experience in technology, including work helping launch monday.com in France. His approach focuses on estimating the cost and resources required for each use case before testing it.
“Companies need to understand the cost of a use case and the resources it will require. The return-on-investment discussion should happen before the pilot, not after.” Zacharie Lahmi, founder of Ellis Partners
Every AI test needs clear goals and a plan for when things go wrong. Lahmi recommends setting measurable targets for usage, maintenance costs, and errors.
“Useful pilots are the ones we really want to see in production. We should agree on success criteria beforehand, so there is no debate after the pilot about whether it worked.” Zacharie Lahmi
Look beyond sales and marketing
Lahmi also urges companies to look beyond sales and marketing when choosing AI projects. While these functions offer benefits that may be easier to link to revenue, he believes everyday operational processes can offer greater opportunities.
“The most boring operational tasks, the core processes of your industry, are the ones you should be thinking about with AI. Ask yourself how you can fundamentally change the way your industry operates, rather than simply maximizing short-term gains.” Zacharie Lahmi
Some enterprise AI projects remain in testing for several months, according to Lahmi. Often, nobody has defined what the project must achieve to justify wider use. He recommends assessing each AI project based on the business problem it would solve.
“Companies should then weigh the expected benefits against the effort required to implement it and the time it would take to see results. A smaller project that delivers results quickly may be a better starting point than a larger one.” Zacharie Lahmi
He suggests examining everyday business operations, too. Improving a process can change how a company works, rather than simply helping employees complete tasks faster.
Size changes the pace
Company size also affects how easily AI projects can be introduced and adjusted. Lahmi notes that smaller businesses often have fewer layers of approval, allowing them to make decisions and change direction more quickly. Large enterprises may face more hurdles because changes can affect multiple teams and existing systems.
Regardless of company size, businesses might need to assess whether an AI system delivers the results they expected. They can track how much time it saves, whether it reduces mistakes, and what it costs to maintain. These measures can help leaders decide whether to continue with a project, make changes, or stop it altogether. A successful demonstration alone might not be enough to justify wider adoption.
From experiment to dependable tool
For stakeholders, the final litmus test for enterprise AI is not just doing the same thing faster, but fundamentally changing how value is created. Launching more pilot projects does not necessarily mean a company is making progress. Each project needs a clear goal and a realistic budget, along with limits on the risks the business is willing to accept.
Companies can review performance after launch and ensure someone has the authority to pause and audit a system if problems arise. These steps can help turn AI experiments into tools that businesses can depend on.