Now in production — Vincu at ProCredit Bank · read the story →

← All programsFree · 60 minutes

The diagnostic, run on SME lending · Book this week

Your SME lending workflow, mapped. Sixty minutes, free.

One call with the team that built and runs AI credit decisioning at ProCredit Bank. You walk us through how an SME application moves today — intake, documents, analysis, committee, disbursement. We tell you which steps AI can take off your analysts' desks, which ones should keep a person on them, and what the first one would cost to put into production.

Book the review Free · 60 minutes · nothing to prepare, nothing to upload
Vincu Decision — a company credit file inside the ProCredit backoffice: AI reasoning, financial indicators, risk grade
Exhibit — Vincu Decision · ProCredit backoffice
<24 hto a micro-loan decision at ProCredit Bank — was about a week
15public registries and records LeadHunter reads per applicant
100%of Vincu decisions logged with their written reasoning
What we map

Most of a week disappears between the steps, not inside them. That is where a machine belongs — and where it does not.

This is the shape we found at ProCredit and the shape we expect to find at your desk. On the call we redraw it in your vocabulary, with your systems in the boxes.

01AI takes it

Intake and completeness

Read the application and its attachments, flag what is missing, and ask the customer for it through a portal instead of a reply-all email thread.

02AI takes it

Registry, court and tax records

Pull the public record on the company and its owners across the official sources — the search LeadHunter already runs across fifteen of them — so the analyst opens a file, not a browser.

03AI takes it

Financial statement extraction

Indicators out of PDFs and scans and into the case, each figure carrying the page it came from.

04AI drafts, a person signs

Risk grade and the written reasoning

The engine proposes a grade and writes its reasoning in full. The analyst approves or overrides — and the override is part of the record.

05AI takes it

The committee pack

The reasoning travels with the decision, so the pack assembles itself from the case instead of being rebuilt in a slide deck.

06A person stays on it

The decision

A human decides. The system makes the decision consistent, fast, and explainable to the committee, the customer and the regulator.

07AI takes it, sealed

Customer communication and documents

Status, requests and signed documents move through a portal and a sealed vault, not attachments — and the customer sees where their application stands.

08Built in

The audit trail

Every decision, route and document event written once and never edited. Hand it to the regulator as-is.

What you leave with
Your SME lending workflow, step by step — the steps worth automating marked, and the ones to leave alone marked too
A rough cost and payback for the first step, whether we build it as a fixed-price Build or your own team does
The constraints we’d expect on the way: explainability for the regulator, GDPR, core-banking integration, adoption in the branch
A straight yes or no on whether AI belongs in this workflow at all
Who should be on the call2–3 people
  • Whoever runs SME lending or credit risk
  • Whoever owns operations — COO, head of operations
  • The CIO, if you want the integration questions answered live

Nobody prepares a deck. Nobody shares data.

How the hour runs
0120 min

You walk us through it

How an SME application moves today: who receives it, what they wait for, where it sits, who signs. The complaints are the map.

0225 min

We probe

Where the documents live, what the core system exposes, who touches a file before committee, what "good" looks like to credit risk and to the customer.

0315 min

We call it

The one step worth doing first, roughly what it costs and returns, and a straight answer on whether there is a case here at all.

Asked by lenders

Does any customer data leave the bank for this?

No. The review is a conversation about how the process runs today — nobody uploads anything. If a Build follows, where the system lives is a scoping decision made with your IT and compliance teams before a line of code is written.

We are a microfinance institution, a leasing company, a factoring desk — not a bank.

The workflow has the same shape: an application, a set of documents, a grade, a decision, and a customer waiting for it. The review is the same hour. Only the vocabulary changes.

Our core banking system is not something you have seen.

Vincu sits beside the core system as the connective layer, on one shared record; it does not replace it. What the core exposes is one of the questions we ask on the call, and it shapes what we recommend first.

What happens after the call?

One of three things. A Roadmap, if several workflows are worth ranking. A Build straight away, if one step is obvious and its data is reachable. Or nothing — and you will have heard that on the call, not in a follow-up email.

Why is it free?

Because the call is also how we decide whether we want to work with you. We turn down work that AI cannot genuinely improve; the review is where that gets decided, on both sides.

Our IT team says they can build this with Claude.

They can probably build the first version, and so can we — we use Claude for it. The review is not about who writes the code. It is about which step is worth automating first, what the regulator will need to see, what the core system exposes, and what it costs to keep running. If the answer is “do it in-house”, we will say so and tell you where the wall is.

One workflow, one hour. Then a price, or a no.

Book the review and bring the person who signs the loans. Or write to the founder directly — he ran the ProCredit engagement and still takes these calls himself.

The next rung, if several workflows qualify

02 · Roadmap

The board-ready plan

€8K–€15K · ~10 days