Transformation
How work with AI reaches the board's quarterly report
The AI Business Center Foundation (AIBC) is a think tank and advisory partner that helps the boards of construction, industrial and technical-services companies adopt artificial intelligence so that it delivers measurable results and builds competitive advantage. “From AI to BI” is our name for the path from the use of artificial intelligence in a company to the measures the board sees every quarter; it describes a direction, not a separate service.
AI is no longer a cost of the IT department. It has become a line the CFO wants to see in figures
Language models produce ever more content. The board needs numbers. Between the two sits the layer we build: names, owners and definitions that make work with AI measurable.
We build this layer using the method of semantic transformation, that is, organising a company’s names, processes and data so that people, language models and board reports understand them the same way. “From AI to BI” describes where this path leads: to a quarterly report in which the board sees what work with AI delivers and what it costs.
BI stands for business intelligence: a company gathers data from its systems, such as sales, finance and production, describes it with shared definitions and presents it to the board as measures in reports and dashboards. In “from AI to BI”, the point is that work with AI does not stop at a sense that employees are saving time, but enters the same report in which the board reviews sales, costs and quality, as a measure with an owner and a definition.
| AI generates | semantics organises | BI accounts |
|---|---|---|
| content, answers, emails, draft documents; the volume grows every month | processes, concepts, data owners and definitions of measures shared across departments and tools | cost, time, quality, compliance and external brand, every quarter, in the board report |
The middle element is what the company pays for. Without it, the growth in content does not turn into a number the board can ask about.
The gap between use and results
According to McKinsey, The State of AI 2026, 8 in 10 employees see time savings from GenAI; only 4 in 10 companies see an impact on operating profit (EBIT). The difference arises in the middle layer: where nobody has described the process, appointed an owner or defined a measure.
Why the letter does not change by itself
The phrase “from AI to BI” sounds like changing a single letter. In practice, between one letter and the other lies work that no tool will do for the company: the same parameter has different names in sales, production and finance, data on tool usage is not assigned to processes, and recovered time has no owner to decide what it should be used for. A dashboard built on such data shows numbers, but does not answer the board’s question.
That is why the sequence is fixed: first competence, then naming the processes, then linking them through a shared vocabulary, and only at the end value measurement. The individual levels are described on the semantic transformation page.
5 groups of measures and their owners
| group | example measures | owner on the company side |
|---|---|---|
| value | process time before and after, throughput, number of cases per person | director of the department running the process |
| cost | licences in actual use, cost per task; tokens only where billing is based on usage | finance, IT |
| quality | corrections, follow-up questions, complaints about content prepared with GenAI | process owner, quality |
| compliance | measures to support staff AI literacy, customer touchpoints subject to transparency obligations, AI use register | compliance, legal |
| external brand | presence, description and recommendations of the brand in language models | marketing, sales |
A full description of the measures and how they are set is on the AI value report for the board page.
What the board sees every quarter
ILLUSTRATION, sample data. The report fits on one page and has five tiles, one for each group of measures:
- Value: time in selected processes compared with the baseline, and the trend over recent quarters.
- Cost: licences in actual use against licences paid for, and cost per task in the processes being measured.
- Quality: corrections and follow-up questions on content prepared with AI.
- Compliance: status of the AI use register and of measures on literacy and transparency, with a list of matters for decision.
- External brand: how language models describe and recommend the company compared with the previous measurement.
Under each tile there is an interpretation and one recommendation for the board. We agree the specific measures and their values with the owners in the company; we deliberately give no figures on this page.
Who does what
| role | who | accountable for |
|---|---|---|
| recipient | board, supervisory board | decisions based on the report |
| measure owner | department director | definition and quality of the measure |
| process translators | department employees | maintaining the descriptions the measures rely on |
| architect and interpreter | AI Business Center Foundation | vocabulary, definitions of measures, brief, interpretation, recommendations |
| BI contractor | the company’s IT or a partner | data model, data flows, dashboard |
We do not implement BI tools. We define what to measure and how. The Foundation is vendor-neutral: we do not sell licences or implementations. We act on the board’s side — we help choose the contractor, prepare requirements and criteria, run the procurement procedure or tender, and assess bids and contractor quality. The board makes the decision.
External semantics in the same report
The same company vocabulary determines whether a model will do the company’s work and whether it will recommend the company correctly to a customer. That is why the “external brand” group goes into the same report as value and cost. We feed it with brand research in language models.
Frequently asked questions
Is this a BI system implementation?
No. We do not implement BI tools. We define what to measure and how, agree the owners of measures and the data sources, and prepare a brief. The data model and dashboard are built by the company's IT department or a partner it appoints, in a tool the company already uses or chooses itself.
From which level of the transformation does this make sense?
The value report makes sense after the second and third levels, once processes are named and linked by a shared vocabulary. Before that, you can report on competence and tool usage, but not on value in processes. The 30-day readiness assessment shows where the company stands today.
Are tokens a good measure of AI cost?
Only where the company pays for usage, for example in integrations and agents billed by consumption. With per-seat licences the company pays for the seat, so the right measures are licences in actual use and cost per task.
How does this work differ from a BI integrator's project?
The integrator builds the data warehouse, data flows and dashboards. We deal with the layer in front of them: names, definitions of measures, owners and which AI-supported processes are worth measuring at all. The two pieces of work complement each other; the integrator receives a brief from us rather than guesswork.
Let’s start with an assessment
The report makes sense once processes are named and linked. The 30-day readiness assessment shows how far the company is from that point. Let’s talk about the assessment.