Skip to content

Where AI earns its keep in lending.

Applications arrive as document packs and leave as decisions. Most of the hours in between go to reading, re-typing and cross-checking. That's the part software should carry.

Workflows

What we'd build first.

01

Document intake

Bank statements, payslips and financials parsed into structured data, with every figure linked back to its page.

02

Affordability and risk signals

Ratios computed the same way every time, with flags for concentration, returned debits and irregular income.

03

Credit memo drafts

A first-draft summary an analyst edits instead of writes, with citations on every claim.

04

Claims triage

Incoming claims sorted, summarised and routed, so assessors start with the full picture.

The pilot metric

Analyst minutes per application, from upload to a reviewable file.

Measured before the Pilot Sprint starts, and again when it ends.

Built with care

Credit decisions stay with people. Systems are built as decision-support under the NCA and POPIA, with an audit trail on every output.

From PXL Labs

Prototype

Mako: AI-assisted credit analysis for SME lenders. Documents in, traceable credit summary out.

See how it works

How we work

Small steps, real numbers.

Every engagement starts small enough to prove itself. Scale it only if the number moves.

  1. 01 · 30 min, free

    Scope

    A call to find the one workflow worth fixing first, and the number that says it worked.

  2. 02 · 2–4 weeks

    Pilot Sprint

    One workflow, one metric, fixed scope. You see the number move before anything scales.

  3. 03 · Scoped

    Build

    The platform or system, built properly: tested, reviewed, documented, handed over.

  4. 04 · Optional retainer

    Run

    Monitoring, improvements and new workflows, month to month.

Have a workflow that should run itself?

Tell us what's eating your team's week. We'll tell you, honestly, whether software can fix it.