Investors & partners · Pre-seed · Warsaw

A working financial workflow. A broader platform opportunity.

LINKIO is developing AI orchestration for financial operations, starting with an internal AI CFO system and a clear path toward external pilots.

Request the Investor Brief Read the milestone gates

What runs today: A single-tenant system reconciling a real settlement register, in daily use.
What is missing: Multi-tenant isolation. The one architectural gap between system and product.
External revenue: None. No paying customers, and no validated price.
Delivery model: One founder plus AI tooling. No engineering team is funded.

The market gap

Horizontal tools solved cards and banks. They left the rest manual.

Where a merchant's money is actually trapped is not the card processor. It is the carrier settlement register, the cash collected at the door and the regional gateway — rails that hold money, release it on their own schedule and report in their own format.

Six systems, six versions of the truth

Storefront, gateway, carrier, bank, accounting and advertising each describe a different slice of the same money.

Three categories, routinely added together

Money that is real, money in transit and money frozen in undelivered orders get summed into a figure that does not exist.

Each rail is configuration, not a rewrite

A settlement register is one instance of a general shape. Supporting the next one is set-up against a shared intermediary.

A sourced, bottom-up addressable count is a deliverable of this round, not a figure asserted on a landing page. No market size is claimed here.

Sequencing

Why AI CFO is the first product rather than one of several.

The pain is measurable

A reconciliation failure has an amount attached. The value of fixing it does not need to be argued in the abstract.

The data proves the engine

Money rails are the hardest case for an orchestration layer: slow, partial, inconsistent, and unforgiving of invented values.

The wedge generalises

What is learned building one rail is reused on the next. Other LINKIO service lines are real upside and are valued at nothing until tested.

Architecture

Three assets exist. One does not — and it is named.

Three of the four components run against a real payment register today. The fourth is what stands between a working internal system and something permitted to hold another company's financial data.

Rail connectors
Read money state from each system on its own terms and on its own schedule.
Built, single-tenantWorking
Ledger and invariants
One classification of money, with assertions that fail loudly rather than drift quietly.
Built, single-tenantWorking
Agent runtime
Diagnostics, approved operations, monitoring and evidence collection.
Built, in daily useWorking
Server-side intermediary
Multi-tenant isolation, per-tenant credentials and an audit trail across tenants.
Not built — the gapRoadmap

Boundary

PetsChoice.Club is the proving ground, not the revenue.

The system runs inside an e-commerce business the founder owns. That gives the company a real settlement register, real delays and real undelivered orders to test against — and gives an investor nothing about willingness to pay.

What it provides

A live payment register with genuine settlement delay
A production environment where the invariants were tested against money that moved
A first integration built and running, not described

What it is not

Not external revenue — it is the founder's own company
Not a customer reference, and never counted as one
Not evidence that an unrelated merchant will pay anything

Path to external pilots

Isolation, then validation, then someone else's data.

01
Data isolation
Per-tenant credentials and separation enforced by a server-side intermediary — the component that does not exist yet.
02
Security validation
An independent third-party assessment, completed and remediated, before any external financial data is touched.
03
External pilots
A small cohort chosen for how their money moves. This is where price, churn and acquisition cost stop being assumptions.

The order cannot be rearranged. Processing an external company's financial data before an independent assessment is not a schedule decision.

Business model

A hypothesis, labelled as one.

Recurring subscription priced by rail, with onboarding and implementation alongside it. Every revenue figure in the model is tagged as an assumption or a target, and a control in the workbook enforces that tagging.

Modelled

Recurring subscription — the price hypothesis this round exists to test
Onboarding fee on each new customer
Implementation, taken by a minority of customers

Carried at zero

Premium tiers — no willingness-to-pay evidence exists
Enterprise contracts — no pipeline; including them would invent traction
Partner and referral distribution — no signed partner
Other LINKIO service lines — real upside, entirely unvalidated

Pricing, unit economics, scenarios, runway and the size and structure of the round are in the brief rather than on this page. They are specific enough to be worth discussing properly, and too specific to publish.

Milestones

Gates with verifiable criteria, in an order that cannot be rearranged.

Each item below is a condition, not an intention. None of them is presented as completed work.

01
IP opinion and trademark clearance
A written opinion on ownership of AI-assisted code, and a completed trademark search.
Verifiable by: the opinion itself, and search results. Due before incorporation.
02
Entity and IP assignment executed
A legal entity exists and the work is assigned to it.
Verifiable by: registry record and executed assignment.
03
Multi-tenant intermediary in production
Per-tenant isolation and credentials running, not designed.
Verifiable by: a second tenant provisioned and isolated under test.
04
Independent security assessment complete
Completed by a third party before any external company's financial data is processed.
Verifiable by: the assessor's report and a remediation record. Sequencing is not negotiable.
05
First external paying customers
Companies unrelated to the founder, paying their own money.
Verifiable by: signed agreements and received payments. This is the event that converts revenue from assumption to fact.
06
Acquisition cost measured on real sales
A cost per customer derived from sales that happened, not an industry proxy.
Verifiable by: spend and win data across the first cohort.

The first two fall due before any round closes. If the opinion on AI-assisted code ownership is negative, the company is not formed — and no investor capital will have been spent finding that out.

Get in touch

Request the investor brief.

The brief covers the cost base, the pricing hypothesis, scenarios, runway, governance and the risks — including the ones that would stop the company. It is shared in conversation rather than published, and there is no download on this page.

This page is not an offer of securities and creates no obligation on either side.
No return, liquidity event or exit valuation is promised anywhere.
Confirmed external paying customers: none.

See the product this rests on →