For financial services firms

Reimagine your B2B lending stack

Embedded financing, credit decisioning, fraud prevention and collections for fintechs, payment providers, marketplaces, lenders and insurers. Designed and deployed by a team that has built them at scale.

Clients

What our clients say

“The team brought deep technical expertise, delivered at pace, and with complete focus on practical implementation, not conceptual slides.”
LG
Lara Gilman
“The Liquidity Lab team have been at the forefront of innovation in AR and AP. No-one knows how to implement technology in this space like them.”
RC
Raphaël Caruso
“Liquidity Lab helped us optimise our DSO in record time. They also allowed us to converge as a Board on the right blend of factoring, credit insurance and in-house collections.”
KD
Ksenia Delaittre
“Liquidity Lab helped us to roll out a new counterparty credit risk framework. The team’s deep technical knowledge combined with their hands-on practical experience made them a great partner for us.”
GM
Gaurav Modgill

Who this is for

Firms launching or scaling a B2B finance proposition

B2B payment providers and fintechs adding pay-later or trade credit. Marketplaces and platforms embedding financing at checkout. Lenders and banks building B2B propositions. Credit insurers, factors and receivables funds digitising origination, underwriting and collections.

  • Built by our team

    100k+

    B2B clients and over 500 million euros financed at the B2B pay-later provider our cofounders built and ran.

  • Fraud reduction delivered

    100x

    Fraud rate cut from 5% to 0.07% at a B2B fintech under active attack, with 98% of buyers approved friction-free.

  • Conversion from instant decisions

    2x

    Sales conversion doubled when trade credit decisions moved from a 48-hour manual check to a real-time engine. Buyers offered terms spent 48% more, with a 3.6x higher repeat purchase rate.

Why work with us

We have built and run these products ourselves

Our cofounders built Hokodo. Our data lead ran fraud analytics globally at Klarna and was Chief Risk Officer at Hokodo. Our technologist was CTO of Hokodo and of the Allianz Trade Digital Agency. We are consultants, not a vendor: we do not sell software, lend or take balance sheet risk, so our advice on what to build, what to buy and who to fund with stays unbiased.

Design

  • Credit policy, underwriting rules, limits and pricing grounded in real payment behaviour data.
  • Funding and risk transfer options: credit insurance, forward flow, factoring and credit funds from our network.
  • Operating model, controls and the business case your board and funders will sign off.

Deliverable: a product and risk design ready to build

Deploy

  • Senior engineers who ship the decision engine, fraud stack and collections agents into your platform.
  • Models and agents that keep learning from your own data after we leave.
  • Your risk and data teams trained and in control of every model and rule.

Deliverable: live components, governed and owned by your teams

What we build with you

Six components, each proven in production

Credit decision engines compared with the traditional approach

Credit decision engines compared with the traditional approach
Liquidity Lab Traditional approach
Data sourceExternal bureau data plus actual payment behaviourExternal bureau data only
Update frequencyReal-time exposure monitoring with automated data feedsQuarterly, based on 6 to 18 month old accounts
Credit limitsDynamic, auto-adjusting to payment behaviourStatic, manually reviewed
Early warningOngoing score refreshes and automated limit adjustmentsYou find out when they stop paying
FocusFix the root cause of long DSO before selling the receivablesFactoring implemented to accelerate cash, overdues still lead to costly financing

How an engagement runs

From design to live components in one to two quarters

  1. 01

    Diagnostic and design

    1 to 4 weeks

    Product, risk and data review on your real transaction data. Target architecture, credit and fraud policy, funding options and a business case with a sized opportunity.

  2. 02

    Pilot

    4 to 8 weeks

    Deploy a first model or agent on a limited segment and measure it against your current process: approval rate, fraud rate, recovery rate, cost per case.

  3. 03

    Scale and embed

    4 to 12 weeks

    Roll out across the portfolio, integrate with your platform and partners, train your risk and data teams, and hand over governed components with named owners and full documentation.

Trusted advisors

  • We are software agnostic. We do not sell a product and never take a commission from vendors or funders we introduce, so our recommendations are unbiased. We help you navigate build versus buy decisions.
  • We focus on outcomes. We know our trade and we are comfortable putting our fees at risk. We get paid when results are delivered.

What makes us different

Three things traditional consulting does not do

Results

Numbers from real engagements

Anonymised at our clients' request. References available on the call.

  • Fraud prevention

    B2B fintech onboarding 4,000 new buyers a month

    0.07%

    fraud rate, down from 5%

    98%
    of buyers approved with zero friction
    > €600k
    of losses stopped within weeks

    Real-time, risk-based verification at point of sale: device fingerprinting, behavioural scoring and step-up identity checks only for high-risk buyers. Adopted as the standard fraud stack for future merchant roll-outs.

  • Automated credit decisions

    B2B services provider, trade credit offered at point of sale

    2x

    sales conversion

    +48%
    average spend
    3.6x
    repeat purchase rate
    4 to 5x
    customer lifetime value

    Instant trade credit approval shown to sales reps during the first order call. 85% of customers are now offered credit immediately, up from a third via a 48-hour manual process. Conversion doubled from 15% to 30%.

  • Agentic collections

    Wholesaler serving 10,000+ retailers a month

    90%

    lower cost per case

    70%
    collection rate on small debts, up from 45%
    8.5 / 10
    customer satisfaction

    AI agents answer customer emails and texts, place outbound calls and handle invoice copies, proofs of delivery and small disputes. On cases above 400 euros they matched the human team at a tenth of the cost.

Questions financial firms ask us

Frequently asked questions

How do you build a B2B credit decision engine?

We start from your data: bureau records, financial statements where available and, most importantly, actual payment behaviour across your book. We build machine learning scorecards on that combined dataset, define limit and pricing rules, and set up real-time exposure monitoring with automated limit adjustments. A first model is typically live on a pilot segment within six weeks, then scaled across the portfolio.

What fraud controls does a B2B BNPL or trade credit product need?

Three layers, applied in proportion to risk: device fingerprinting and network signals at onboarding, behavioural scoring on the order and buyer profile, and step-up identity verification only for the high-risk minority. Done well, over 98% of good buyers pass with zero friction. At one B2B fintech onboarding 4,000 buyers a month, this cut the fraud rate from 5% to 0.07% within weeks.

How long does it take to launch an embedded financing product?

A diagnostic and design phase of one to four weeks defines the credit policy, underwriting, pricing, funding and operating model. A pilot on a limited segment follows and runs for four to eight weeks, then roll-out and integration take four to twelve weeks. From a standing start, a controlled launch is realistic within one to two quarters, depending on funding arrangements and integration complexity.

Should we build or buy our collections and decisioning stack?

It depends on volumes, differentiation and your engineering capacity, and we have no stake in the answer: we sell no software and take no vendor commissions. We map the vendor landscape (we publish an AI collections vendor map), define the target architecture, and run the selection if buying makes sense. Where building wins, our engineers deliver the models and agents alongside your team.

Can AI really handle collections conversations?

For the routine 80% of an AR inbox, yes. Requests for invoice copies, payment promises, proofs of delivery, statement queries and small disputes follow deterministic playbooks. Tool-calling agents resolve these instantly, with full audit trails, and escalate anything that needs judgement. At a wholesaler serving 10,000+ retailers, agents matched human recovery on cases above 400 euros at a tenth of the cost and lifted the collection rate on small debts from 45% to 70%.

Factoring, credit insurance or in-house collections: which is right for us?

Usually a blend, and the right mix depends on your customer concentration, margins, growth plans and cost of capital. Factoring accelerates cash but costs more when your ageing is poor. Credit insurance protects against large losses. Strong in-house or AI-assisted collections improve both. We have no product to sell, so we model the options on your own receivables data and help your board converge on the right combination, as we did with Markit.

How do you work with our existing risk and data teams?

As an extension of them. Your risk team owns the policy and your data team owns the platform. We bring specialist credit, fraud and collections experience and senior engineers who ship. We work in your repositories and tooling where possible, document everything, and train your people so that models and agents keep improving after the engagement ends.

Do you take any balance sheet or origination risk yourselves?

No. Liquidity Lab is a pure consulting firm. We do not lend, insure or factor, and we do not sell software. Where funding or risk transfer is needed, we introduce and structure arrangements with our network of credit insurers, factors, credit funds and brokers, and we help you negotiate terms. Our fees are consulting fees, and in the pilot and roll-out phases they are paid only when agreed results are delivered.

Contact

Let's build your B2B finance stack.

Book a 45-minute introductory call. We will come with questions about your situation and leave you with a first view of the opportunity.

  • Product and risk review scoped in the first call
  • Build versus buy view on your current stack
  • Introductions to our funding and risk transfer network

We reply within one working day. No newsletters, no automated sequences.