Resources
Artificial intelligence applied to business decisions
Deciding on an artificial intelligence project means framing a use case from a business problem, estimating its full cost and measuring real usage. Those three skills belong to management, not to engineering.
Start from the problem, never from the technology
A solid use case is stated like this: this decision is taken today in this way, it costs this much time, and automation would make it faster or more accurate. The technology comes afterwards, as an answer. That inversion rules out projects starting from a tool in search of a use. It also makes costing possible, since value is measured on the existing decision, using indicators the organisation already tracks.
- An existing decision, described by who takes it and how often
- An expected gain expressed in the unit the board already tracks
- A full cost: data, integration, training, operations
- A stopping criterion set before launch, quantified
- An internal usage policy, written and circulated
The certificate for decision-makers
Four blocks, no technical prerequisite. The first establishes what a model actually does, the last installs governance: that order is what makes trade-offs sustainable.
| Teaching block | Year | Hours | Credits |
|---|---|---|---|
| Understanding without coding | 1 | 30 h | 3 |
| Identifying use cases | 1 | 30 h | 3 |
| Generative assistants at work | 1 | 30 h | 3 |
| Governance and compliance | 1 | 30 h | 3 |
Four real-usage indicators
Frequency of use
A tool opened daily by the same team says adoption. Accounts created says deployment.
Time returned
The decision's duration before and after, measured on the same files: the gain that stands up in committee.
Output quality
A sample check, sustained over time. It is part of the project, just like the tool itself.
Running cost
Monthly cost per decision handled. It is what decides the project's next phase.
Questions from executives
Do I need technical skills to decide?
The decision-makers' certificate opens with no technical prerequisite: it covers framing, estimation and governance. It runs over 4 months at 12 hours a week, while you keep your job.
How do I estimate a project's full cost?
By adding four items that are often not counted together: data preparation, integration into the existing process, team training and monthly running costs. The framing block works that calculation on a real case brought by the participant.
Where does responsibility for the outputs sit?
With the person who decides, and the internal usage policy states it plainly: which decisions go through a human check, who checks, and on what sample. The governance block builds that document with you.
Explore next
- Preparing dataSources, formats, missing values, duplicates, reference data and traceability: the data preparation method that makes an analysis defensible.
- DashboardsThe starting question, chart choice, visual hierarchy, thresholds and alerts, distribution: building a dashboard that triggers decisions.
- Misleading analysisSelection effects, chosen denominators, correlation versus causation, truncated axes: the four checks to run before accepting a figure.
- Machine learningRegression, classification, cross-validation, model selection and interpretability: what a model actually does and how it is validated.
- Models in productionInference service, drift monitoring, retraining, running cost and the trade-off between local and cloud processing.
- Data governanceData ownership, life cycle, hosting, retention, reversibility and vendor dependence: framing governance.