A practical take on Kit Digital and AI

Kit Digital and Kit Consulting are not a badge for the website. They are a frame: diagnose, plan, implement, and leave the team able to continue. AI fits when it is a means to a process — not the title of the slide deck.

On the first call you should not be asked for “your AI vision”. You should be asked for a flow, a volume, an owner and a number you already care about. If the supplier talks only about models and never about invoices, tickets or stock, you are in a demo.

This is the orientation we use in Barcelona with SMEs in any sector: translate the catalogue to your size, do not sell the trendiest category.

What orientation is (and is not)

It is translating the catalogue to your size: ten people do not implement what one hundred and twenty implement. It is picking a category you can evidence, not the modern-sounding one. It is writing down the delivery, the operator and the 30- and 90-day metric.

It is not stuffing the memo with jargon, promising a chatbot that “does everything”, or vanishing until audit time. It is also not “we will build the site and then there is AI”: if the habit does not stay, the reviewer sees an orphan deliverable.

Where AI sits in the catalogue

• Advisory (Kit Consulting). Data inventory, use-case priority, GDPR risk, training plan, supplier criteria.
• Site and capture (Kit Digital). Cited FAQs, on-site help, forms that keep the lead, honest measurement.
• Operations. Document extraction, reply drafts, matching, alerts — always with a human check.
• Training. On your stack, with champions per area and a review at two and six weeks.

AI is almost never its own showcase category. It is a capability inside site, operations, management or training.

Questions to ask the digitalising agent

• What is delivered in week 2, not only week 12?
• Who on my team runs this the Monday after training?
• Which data does it touch and where does it live?
• What do we switch off if the error is worse than the saving?
• How does success show up in a number I already watch?
• What happens if Red.es asks for evidence of use, not only of installation?

How bonuses get burned

Buying tools before mapping the process. Training everyone in one afternoon with no protected time. Measuring workshop smile-sheets instead of use. Opening three categories at once. Signing a scope only the supplier understands.

The reviewer does not reward originality. They reward coherence: diagnose → plan → implement → habit. AI is a line in that thread.

An order that usually fits a quarter

Inventory and priority on one page. Then one-flow pilot. Legal basis in parallel. Train the two people who will touch it most. Numbers at day 30. Only then a second flow.

That order fits a quarter of real work. It does not fit a committee that wants “AI everywhere” for the adhesion photo. The photo does not justify; the habit does.

Bring this to the assessment

• Sector, headcount, current tools.
• One repetitive task with owner and volume.
• One constraint: budget, deadline, sensitive data, or an untouchable system.
• One metric that already exists, even if crude.
• If you already have a voucher or call in mind, the category you are looking at — so we can say whether it fits or you are forcing it.

What we do not promise

We do not promise adhesion: Red.es decides that. We do not promise that “any category works for AI”. We do not promise a public chatbot in month one. We promise a plan you can run and justify, with training that survives week one.

What a useful first call looks like

Length: 30–40 minutes. You bring sector, headcount, one repetitive task and one constraint. We return: whether the case fits Kit Consulting, Kit Digital, or a private pilot; which checklist box is missing; and whether it is worth waiting for a datum or an owner.

What we do not do on that call: demo a model, promise adhesion, or recite the whole catalogue. If you hang up without a pilot sentence or a clear “not yet”, the call failed. Ask for it again, shorter.

If a vendor is already on top of you, bring the scope and price in writing. Reviewing someone else’s quote is part of orientation, not a courtesy extra.

A note on timing and the voucher calendar

Public calls have windows. That pressure is real and it is also how bad scope gets signed. If the window is three weeks away and you have no owner, spend those weeks on the inventory and the sentence, not on picking a logo.

A late, coherent file beats an on-time file that describes a chatbot nobody will run. Reviewers have seen both. So have we.

Frequently asked questions

Do we need to be adhered to talk?

No. An honest diagnosis is useful even if the voucher does not fit yet.

Is AI its own category?

Almost never. It is a capability inside site, operations, management or training.

How many flows at once?

One measured flow. The second is earned by the first number.

Can you write the whole memo for us?

We can draft from your evidence. We do not invent processes the team does not run. Reviewers notice.

Next: request an assessment or read Kit Consulting and Kit Digital. Related: where to start with AI and realistic budgets.