Cases

Where AI genuinely changes operations, and what a person still decides.

These are examples of situations I usually step into, not client cases. They all follow the same pattern: AI prepares, a person reviews, decides and answers for it.

Clients I've worked with

Levante UD

Executive leadership

«I have the data, but I can't tell how things are going»

Example

The leadership team's memory

Now: Something is decided in the leadership meeting and by March nobody remembers why or who owned it.
I do: Every meeting yields decisions and owned tasks, in a memory that doesn't depend on who attended.
You: You approve what becomes company policy.

Example

The leadership report

Now: Each department builds its report by hand, with figures that don't match each other.
I do: One up-to-date view, pulled straight from the systems, with nobody preparing it.
You: The leadership team debates decisions, not numbers.

Example

The CRM nobody fills in

Now: A large sales team, incomplete records and a pipeline nobody believes.
I do: It fills itself from emails and meetings. I don't ask sales for more discipline. I take admin work off them.
You: You see the real pipeline without chasing anyone.

Example

Asking your data

Now: "Which customers have invoices overdue by 60+ days?" takes two days and three emails.
I do: An assistant that queries your systems with each person's permissions.
You: You ask in plain language.

Example

Junior work that disappears

Now: AI absorbs entry-level tasks and nobody has decided how tomorrow's seniors will be trained.
I do: I design which tasks to automate and which to keep as training, with people supervising the machine.
You: You decide which work is training and which is cost.

Finance and operations

«Month-end close still runs on overtime»

Example

Supplier invoices at scale

Now: Thousands of invoices a month, from hundreds of suppliers across several entities, checked by hand.
I do: They are read, matched against order and delivery note, and only the mismatches reach a person.
You: Your team handles the exceptions.

Example

Multi-entity reconciliation

Now: Every close, matching bank and intercompany entries across several entities by hand.
I do: The reconciliation proposal arrives ready, with every difference explained.
You: A person reviews and confirms.

Example

Orders arriving as PDFs

Now: Customers send orders by email and a whole team retypes them into the ERP.
I do: A draft order in the ERP with the lines that don't match flagged.
You: You decide on the doubtful ones and confirm.

Example

Carrier and supplier claims

Now: Incidents handled in endless email threads, and money that never gets claimed.
I do: Every shipment is watched and the claim is drafted with the paperwork attached.
You: You review and send.

Example

Data from statements and contracts

Now: Analysts spending most of the day copying figures out of documents.
I do: Documents are read and figures arrive structured, each linked to its source.
You: Analysts analyse and review what looks off.

Example

Standard contracts

Now: Every contract is drafted from an old template and checked by hand in legal.
I do: A form of variables generates the draft from your past contracts.
You: Legal reviews and signs. The legal judgement stays theirs.

Technology and transformation

«We have twenty AI pilots and none in production»

Example

Pilots that never reach production

Now: Twenty AI pilots open, none in production and nobody deciding which to close.
I do: I review each one: value, data, risk and running cost. I recommend closing most and taking two or three to production.
You: The leadership team decides with the reasoning in writing.

Example

Copilots nobody uses

Now: AI assistant licences for the whole workforce, and almost nobody knows what to ask it.
I do: I connect the assistant to each role's context: its documents, its systems and its repeated tasks.
You: You measure real use, not licences.

Example

AI and the ERP

Now: The models are powerful, but they can't see the ERP, the CRM or the documents where the business lives.
I do: I build the plumbing: connections, permissions and context so AI works on real data.
You: You decide what data each agent sees.

Example

Governing AI agents

Now: Either the AI does nothing, or it asks for a click on everything until people approve without looking.
I do: I design what it decides alone, what it asks and what always belongs to a person, with every decision logged.
You: You govern instead of operate.

Example

From prototype to production

Now: Something built with AI in two weeks that nobody dares open to customers or auditors.
I do: I review what's missing: data, permissions, security, compliance and who maintains it.
You: You decide what gets hardened and what gets dropped.

Example

Integrations nobody wants to touch

Now: Payments, channels, messaging and invoicing held together with patches only one person understands.
I do: Maintained, documented connections that alert you when something breaks.
You: You find out before your customer does.

Recognise yourself in one?

The free diagnosis tells you which ones are worth it in your case, and which aren't.

Get a free diagnosis