Skip to content
Black Wings
es

AI in business processes

AI is not a chatbot on your website.It's the work you stop doing by hand.

We put language and vision models inside your processes: reading documents, classifying, extracting data, deciding. Wherever someone is copying from a PDF into a spreadsheet today — that is where AI actually earns its keep.

The problem

Does any of this sound familiar?

  • Someone types data out of documents

    Invoices, delivery notes, receipts, orders. A person opens them one by one and transcribes them. It's slow, it's expensive and it goes wrong — not because they're bad at their job, but because the job is like that.

  • You tried an AI tool and nothing changed

    You bought a licence, people used it for two weeks and went back to the old way. Of course they did: an AI that lives outside your process is just one more tab to open.

  • You have the data but not the answers

    Your ERP has stored years of information and you still don't know which product actually makes you money. The data exists; what's missing is someone to read it and tell you.

What we do

What we do, specifically

  • Automatic document reading

    Vision models that open the PDF or the photo, work out what it is and pull the fields you need: amounts, dates, taxable base, supplier. With validation afterwards, because a badly read figure that goes in unchecked is worse than having nothing.

  • Whole processes, automated

    Not one isolated step: the full chain. The document arrives, is read, is checked against your system, is recorded, and whoever needs to review it gets told. The person moves from typing to supervising.

  • AI inside your software, not beside it

    We build it into the tools you already use — your ERP, your back office, your internal app — so nobody has to switch context or learn a new tool.

  • Cost and error under control

    Every model call costs money and can get things wrong. We measure both from day one and put the limits and the checks in place before anything reaches production.

The proof

Anotados reads receipts with AI every day

In Anotados the user photographs a receipt and the system returns the supplier, the date, the taxable base, the VAT and the total, applying Spanish and Portuguese tax rules. It is in production, charging through Stripe, with companies using it. It is not a pilot we wheel out for demos: it is the product we live off, which is exactly why we know where it hurts.

See Anotados

How we start

How we start

  1. 01

    We look at the process, not the technology

    One session to see what is done by hand today, how many times a day, and how long it takes. Without that, any quote is a made-up number.

  2. 02

    We pick the part that hurts most

    We start with one concrete, measurable process, not with "rolling out AI". If we can't tell you how many hours it saves, we don't do it.

  3. 03

    A trial on your real documents

    Before building anything, we run your own documents through the model and show you the output and the hit rate. Your data, not a tidy example.

  4. 04

    A written proposal

    Scope, timeline and price agreed, with what's in and what isn't. From there it's your call.

Everything starts with an analysis and ends in a written proposal, with the scope and the timeline agreed up front. No automatic sign-up and no catalogue price: every company has a different problem, and a real quote means looking at it first.

Which process is eating your hours?

Tell us and we'll say whether AI solves it — or whether it doesn't.

Let's talk