Most companies investing in artificial intelligence today do not have a budget problem: they have a return problem. There are projects launched, licences contracted and expectations created, but the business case remains unproven. That is why AI project guidance has become the factor that weighs most heavily on the final outcome: not the chosen model or the platform, but who helps you connect the technology to a measurable business objective and sustain that connection over time.
What is interesting is that this conclusion does not come from a marketing campaign, but from the data that the vendors themselves are putting on the table.
What the data says about AI projects that fail
In an article published on the SAP News Center (news.sap.com), SAP compiles several references worth reading carefully, regardless of the brand:
- RAND Corporation estimates that 80% of enterprise AI initiatives fail to deliver the expected business value, twice the failure rate of a conventional IT project.
- S&P Global Market Intelligence found that 42% of companies abandoned most of their AI initiatives in 2025, compared with 17% the previous year.
- The MIT GenAI Divide report notes that most organisations investing in generative AI see no measurable return on their bottom line, and that having an experienced external partner practically doubles the success rate in deployments.
The underlying reading is the same one we have been seeing in the field for years: technology is rarely the bottleneck. What separates a project that transforms the business from one that quietly fades away is guided execution: someone who translates a product roadmap into a concrete result for the company.
Why AI project guidance makes the difference
When an SME starts with AI without guidance, the same pattern tends to repeat itself. First the tool is chosen and then the problem is sought. A flashy pilot is run that gets everyone excited in a demo, but which nobody knows how to integrate into day-to-day work. And when the time comes to scale, the questions that should have been raised at the outset appear: where does the data come from? who validates it? who answers if the system gets it wrong? how much does this cost when a hundred people use it instead of five?
Serious guidance reverses the order. It starts with the process that hurts —the month-end close that drags on, the delivery note that is keyed in three times, the quote that takes two days to produce— and only then decides whether AI is the answer. Sometimes it is. Other times, what is missing is a properly configured ERP or a simple automation with Power Automate. Saying that is also part of the job of a good technology partner.
Data first, AI afterwards
No AI performs well on disorganised data. If customer, stock or cost information lives scattered across the ERP, five spreadsheets and one particular person’s inbox, any model will give poor or outright incorrect answers. That is why, at Tisa, AI work almost always starts in the same place: identifying which datasets are genuinely relevant, organising them and establishing data governance protocols aligned with the applicable regulations. It is the least glamorous part of the project and the one that most determines the outcome.
A measurable use case, not a showcase
The second piece is choosing the first battle well. An AI use case must be answerable with a number: how many hours are saved per month, how many errors are avoided, how many days a cycle is shortened. If there is no metric, there is no project, there is a demonstration. Working with a scoped prototype makes it possible to validate the idea and reduce risk before committing to large-scale investment.
The ERP as the foundation: why Business Central changes the equation
This is where the discussion needs to come down to earth. For an SME or a mid-sized company, the natural place where AI starts to make sense is not a separate laboratory: it is the system where operations already live. Microsoft Dynamics 365 Business Central covers finance, purchasing, inventory, sales, manufacturing, projects and services in a single system; in other words, it concentrates exactly the data that AI needs in order to be useful.
On that foundation, the Microsoft ecosystem adds layers that fit without forced integrations: Power BI for dashboards, Power Automate to automate administrative processes and Copilot as assistance built into everyday tools.
The message for a business leader is simple: every improvement in the ERP —cleaner data, more consistent processes, fewer manual tasks— is also an investment in the future ability to apply AI meaningfully. They are not two separate projects; they are two phases of the same journey.
Continuous transformation, not a project with an end date
Another idea that the SAP article highlights and that we share: transformation is not a milestone, it is a journey. An ERP implemented ten years ago and never reviewed ends up holding the business back just as much as a pilot AI that nobody maintains. Value appears when there is someone who periodically reviews what works, what is superfluous and what new possibilities are worth pursuing.
That is the kind of relationship we have maintained at Tisa since 1987: a Microsoft partner and ISV, with our own verticals on Business Central for hospitality, retail and POS, distribution, sports centres and fleet management, and experience in sectors such as construction, the food industry and commerce. You can see examples in our success stories.
Where to start
If you are considering an AI project, three questions before signing anything:
- Which specific process do I want to improve and how will I measure the result?
- Is my data in a fit state to support that use case?
- Do I have someone to support me beyond go-live?
If any answer is unclear, let’s talk. At Tisa we offer a no-obligation assessment to analyse your situation and tell you honestly whether AI is your next step or whether it is better to put your house in order first. Write to us at info@grupotisa.com or call us on (+34) 971 305 885.
Because AI project guidance is not an extra in the contract: it is the part that decides whether the investment turns into results.
Reference source: SAP News Center, «AI and Transformation Guidance for Every Customer» (November 2026).
