AI & Digital Transformation

Each of the three projects shown here starts from a specific, repetitive and measurable task: replying to Google reviews for six car brands, entering supplier invoices, reconciling bank deposits in Oracle EBS. Before choosing a technology, we measure the volume handled, the time spent and the error rate. Those figures then serve as the baseline for judging the result.

Generative AI, AI agent or RPA: which tool for which task?

An RPA robot suits fixed rules applied in the same order every time, such as a bank reconciliation. An AI agent takes over when documents change format from one supplier to the next. Generative AI drafts replies tailored to each message, a customer review for example. These building blocks are often combined: the language model writes, the robot publishes or files.

Who are these projects for?

Finance departments, customer service teams and distribution networks that process a large volume of documents or messages every day. An SME can apply the same method on a smaller scale: one process, a limited scope, then a wider rollout once the gain is confirmed.

What each case study covers

The starting point, the chosen solution, the rollout steps and the results measured after go-live. Enough to compare the situation with your own: volume, existing tools, teams involved.

What to check before starting a project

The quality of the input data, access to the tools involved (ERP, mailbox, review platform), the human approval steps and data protection. These points are settled during scoping. Discovered midway through a project, they delay go-live.

Further reading

• AI in business: 5 quick-ROI use cases for SMEs
• Automation: what you can hand over to a machine (and what you can't)
• How do you audit your organisation's AI maturity?

Does a task take up several hours of your team's week? The free AI audit checks whether it can be automated, and with which tool.