Core systems and AI agents for banks, card issuers and lenders

Own your core systems. Compound your edge.

Xenomi builds card, lending and ledger systems, and the AI agents that finish the work around them. Start with both, with the core alone, or with agents on the core you already run. In every case the source code lives in your repository, runs on your infrastructure and is yours to change without us.

Book an architecture review

A 30-minute call with a founder. Leave with a build-vs-rent view of one system.

Watch one job, start to finish
Three ways to start
  1. 01 · Core + AIA new core, with its agents.

    We build the card, lending or ledger system around your rules, with the gateway, agents and evals from day one.

  2. 02 · Core onlyA new core, on its own.

    The same system and its evals, without agents. Add them when you choose.

  3. 03 · AI on your coreAgents on the core you run.

    Your current systems stay as they are. We add the gateway, agents and evals on top.

In all threeEvals on every change · the source is yours
What agents change

Remove the handoffs. Keep the controls.

Much of the work around a card or lending system exists because software could not carry context from one system to the next, so people did. Agents can. What no agent can do is reason away a missing settlement file, a balance that changed mid-request or an action nobody authorized.

  1. Goes

    People carrying work between systems.

    Copying a reference from one screen to another. Re-keying what a customer already said. Typing in what a bureau already holds. Agents take this over.

  2. Agents investigate

    Someone reading unclear evidence.

    Why a total differs. Whether an alert is real. Agents test each explanation and resolve what the evidence supports. The rest reaches your team as a prepared case.

  3. Stays

    A control, or a wait.

    An adjustment needs authority. A settlement waits for the bank. These stay. The system enforces them and records that it did. What goes is the data entry wrapped around them.

What your team sees is what is left: approvals, new kinds of exception and changes to policy.

One authorization
Beforeagents work here
    Insideno agent in this path

      Afteragents work here
        Illustrative timings0100 request · 0110 response
        One job, start to finish

        A break, investigated to the end.

        A settlement total differs from what your books expected. The job starts when the file arrives, before anyone opens a queue. It ends as a break closed with evidence, or as one specific decision for your team.

        What the job has established

          Terms in this job: presentmentpostingbreakdry run

          Reconciliation job
          running
          Illustrative jobClosed only when the ledger agrees

          We count verified resolutions and reopened cases. A high closure rate on its own can hide bad decisions.

          Their agent. Your agent. Your rules.

          The agent asks. The system of record decides.

          Your customers will bring their agents, and you will build your own. Either way, a request made in words becomes one named operation. It is checked against who the agent acts for and what it may do, tried before it runs, and executed once.

          Agent gateway permitted
          Illustrative requestsWorks on systems we build and systems you already run
          • Authority is a stored record.

            Who an agent acts for, for what purpose, up to what limit and until when. It is checked on every call and withdrawn in one place.

          • Consent comes from the customer.

            You capture it, with its purpose and its date. An agent can carry a request for consent. It cannot give the answer.

          • Every action is logged.

            What was asked, the operation it became, what was checked and what the system recorded. Your auditors read the same log your engineers do.

          What we build

          Each system has a part that stays strict and a part that changes.

          Build the card, lending or ledger system you need, or connect agents to the ones you already run. Select any box on the map to see what nobody gets to change, what agents take over and one failure we design for.

          Every box opens. Select one.

          Your infrastructure · your repository

          Cards

          Lending

          Books

          Yours: built around your rules, or already runningXenomi’s: licensed to you as sourceOutside your wallsCore engines under your selection
          Xenomi Core

          The hard parts, built once. Yours to run and change.

          Every card, lending and accounting system has two parts. One is the same at every institution, because card networks, regulators and arithmetic set its rules. The other is different at yours, because you set them. We build the first part once as Xenomi Core, prove it with evals and license it into your system as source code.

          Core engine · cards

          Transaction processing

          Card network messaging, authorization, reversal and clearing. Every financial event carries one lasting identity, so a retried request never becomes a second debit and a late reversal finds the right transaction.

          Try it: lose a response
          With one lasting identity
          50,000.00available · 0 holds

          Without it
          47,500.00available · 1 hold

          Core engine · lending

          Loan lifecycle

          What must be true before money moves, and the states a loan passes through after it does. Each requirement is a recorded fact with its source, its date and how long it stays valid. Your credit policy makes each decision and stays yours.

          Try it: supply the facts in any order
          blocked

          Core engine · cards and lending

          Accounting

          Double-entry books in exact amounts, and the calculations that decide what they record: interest, fees and taxes, statements, and the order in which a payment is applied. Any balance can be traced back to its entries, and a posting can be tried before it is made.

          Try it: backdate a payment
          Posted on the 17th · due on the 14th · 25,000.00 paid · interest at 36% a year

          Core engine · every system

          Security and cyber defense

          Who may do what, for every person, service and agent, and keys that stay in your hardware security module. Attackers now use AI to probe every system at once, so the defense does too. Defense agents investigate each signal to a conclusion and contain an attack inside limits you set in advance.

          Try it: an attack, contained
          Defended

          Undefended

          What a defense agent may do
          On its own

          Narrow, timed blocks and holds

          Needs your team

          Reissuing a card, reversing a posting

          Never

          Touching a key, switching off a control

          Illustrative attacksEvery action above is logged
          • Built once by Xenomi

            The Core is the same everywhere.

            It follows rules nobody gets to choose: card network specifications, accounting principles and cryptographic standards. Xenomi builds it once, owns it and updates it when those rules change. The agent layer and the evals are part of it.

          • Set by you

            Your rules are yours alone.

            Your products, prices, policies and partners, held as versioned configuration. A change is a diff, a run against past activity and an approval, whether your team writes it or an agent proposes it.

          • Held by you

            You hold the source to both.

            All of it runs from source code in your repository, on your infrastructure. Your engineers can read and change every line, and take new Core releases when you want them.

          Evals

          Change anything. Know it still works.

          Evals are test packs for your systems, your agents and the jobs they finish. They run before go-live and on every change after it, so your team can update the system without us and trust the result.

          • E

            For your systems.

            Scenarios taken from real operations: the duplicate request, the reversal that arrives late, the settlement file that never arrives. Each one states what must be true about the money afterwards.

          • A

            For your agents.

            Each agent is tested on the jobs it has to do, the requests it must refuse and the approvals it must wait for. What an agent must refuse gets as much care as what it must do.

          • J

            For finished jobs.

            A job that reports success is checked against the records: the break closed without a cause, the decision made on stale evidence, the job interrupted after its posting had already succeeded.

          • +

            For what you already run.

            We also run evals against systems and agents we did not build, so you can see where you stand before you change anything.

          Eval run release held
          What changed
          Illustrative runEvery run is kept for your risk team and auditors
          Why own it

          Stop renting your core.

          You want to launch a product, change a lending rule or fix a customer problem. You should not have to wait for a vendor’s schedule, pay more every time you grow, or keep renting screens and queues for work that agents are taking over.

          Your edge is becoming software.

          You have spent years learning who to lend to, how to price risk and when to step in. That know-how is becoming lending rules, agent workflows and AI that serves your customers. When a vendor controls the systems or the AI, you learn and then wait to act.

          Know-how

          • Who to lend to
          • How to price risk
          • When to step in

          Software

          • Lending rules
          • Agent workflows
          • AI that serves your customers

          Keep it inside your walls.

          Sending customer data, business rules and proprietary logic to external AI services, without clear controls on access, storage and reuse, can expose the know-how that makes your business valuable. We run models and agents on your premises or in your private cloud.

          What you need to own is also getting smaller. Much of what a rented system charges for is screens and queues. As agents take over that work, the part worth owning is the part that records what happened to the money, and the rules for who may change it.

          Rent or own: find your crossover In multiples of what you pay in rent today
          Rent is charged per transaction, account or loan, so it grows with you.
          A build fee for the agreed scope, paid once.
          Infrastructure and the people who run the system.

          You keep paying rent through year one, while you build and migrate. Your own running costs grow at a quarter of your volume growth. Card network and third-party fees are the same on both paths and are left out.

          Crossover in 3.2 years

          Keep renting Build and own
          Cumulative cost over seven years of renting compared with building and owning
          Show the numbers
          Illustrative modelWe redo it with your volumes and your plans in the architecture review

          Pay for the build. Keep the system.

          A build fee for the agreed scope. No Xenomi fee per transaction. No Xenomi license that grows with volume.

          You still budget for infrastructure, card network fees, third-party services and the people who run the system. We compare those costs with what you pay to rent, using your volumes and your plans.

          What you pay
          Build feeagreed scope
          Fee per transactionnone
          License that grows with volumenone
          What you get
          Full source code, in your repositoryincluded
          The right to change it without usincluded
          Evals, documentation and trainingincluded
          How we work

          We build it with you. You take it from there.

          Your team works alongside ours from the start. The code, the knowledge and the ability to keep building all come with the system.

          1. 01

            Blueprint

            Choose one system or one recurring job, such as a class of reconciliation breaks or disputes. Map what you have, compare building with renting, and agree the scope, the cost and how a correct result will be measured.

          2. 02

            Build

            Assemble the system from Xenomi Core and build what is specific to you, in your repository. Deploy on your infrastructure, with your team involved throughout.

          3. 03

            Test and connect

            Run the evals on normal operations, duplicate requests and recovery from failures. Complete the integrations and the certification work the system requires.

          4. 04

            Run together

            Operate alongside your engineers. Handle exceptions together and train your team on the system in production.

          5. 05

            Take it over

            Your team takes over with the source code, evals, documentation and runbooks. Change the system without us, and keep us for support and Core updates when you want them.

          Agents earn their limits one rung at a time.

          No agent starts by acting on live money. It starts on your past cases, then watches live work, then prepares cases for people, and only then acts alone on small, reversible operations.

          Each rung has a measure that has to hold before the next one opens.

          Our experience

          We know how core systems and platforms are built. We bring the financial systems experience and the AI engineering into one team.

          Our team’s experience spans multiple decades in payments and lending: secured and unsecured lending, credit cards, transaction processing, reconciliation, card network certification, authentication and tokenization, built in earlier operating roles.

          That experience shapes how we build AI for financial services: around real accounts, real money and the people responsible for both.

          • Secured and unsecured lending
          • Credit cards
          • Transaction processing
          • Reconciliation
          • Card network certification
          • Authentication and tokenization

          Own what comes next.

          Bring us a card, lending or ledger system you want to run in-house, a recurring job you want agents to finish, or a system you want tested. We’ll work out what it takes.

          Book an architecture review

          A 30-minute call with a founder. Leave with a build-vs-rent view of one system.