AI Financial Intelligence
Models read every transaction as it happens — surfacing anomalies, forecasting cash position, and flagging exposure before it lands on a month-end report.
We combine artificial intelligence, blockchain, and modern financial infrastructure to make payments, lending, and financial operations faster, smarter, and more accessible.
We are an applied AI and financial infrastructure company operating in stealth. Rather than launch early and iterate in public, we are building the rails, risk models, and controls first — because the businesses we serve cannot afford a beta.
Indonesia is the fourth-largest country in the world and one of the fastest-digitizing economies on the planet. Yet the financial plumbing underneath it — reconciliation, settlement, credit decisioning, treasury — still runs on batch files, manual review, and systems that were never designed to talk to each other.
We are building a single, intelligent financial layer for that economy: one integration that moves money across local rails, one set of APIs for payments, lending, and embedded finance, and a model layer that reads every transaction so risk and reconciliation stop being someone's afternoon.
The team comes from payments engineering, credit risk, and large-scale distributed systems. We are currently working with a small number of design partners under NDA ahead of a public launch.
Give every Indonesian business the same financial infrastructure the largest institutions take for granted.
A warung in Bandung and a treasury desk in Jakarta should reach the same rails, the same settlement speed, and the same credit intelligence — without an enterprise contract in between.
Fraud, credit, and reconciliation decisions belong inside the transaction — measured in milliseconds, explained in plain language, and auditable months later.
Data residency, auditability, and supervisory alignment are architectural decisions. We made them at the beginning, because they cannot be retrofitted.
Indonesian businesses now accept money through a dozen channels — QRIS, virtual accounts, e-wallets, cards, direct transfer, cash on delivery. What they don't have is one place where all of that becomes a single, trustworthy view of their money.
Every channel has its own settlement window, fee structure, dispute flow, and file format. Finance teams rebuild the same reconciliation spreadsheet every month because no system holds the whole picture.
Millions of businesses generate rich, real-time transaction data and still get assessed on collateral and photocopied statements. Creditworthy operators are declined because the data that proves them is unreadable to the underwriter.
Shipping a payment feature means separate contracts, sandboxes, and failure semantics per provider. Engineering time that should go into product goes into keeping seven connections alive.
Money movement, intelligence, and trust are not separate products stitched together after the fact. They are three layers of the same system, and each one assumes the other two exist.
A single API across Indonesian rails — QRIS, BI-FAST, virtual accounts, e-wallets, cards, and direct transfer — with consistent idempotency, retry, and dispute semantics no matter which rail carries the transaction.
Models that sit inline with the flow of funds: scoring risk before authorization, underwriting from real cash-flow behaviour instead of collateral, and matching settlement to invoice automatically. Every output carries the reasons behind it.
A distributed-ledger settlement record gives every participant one shared, tamper-evident version of what happened — cryptographic proof for reconciliation, disputes, and audit, without asking anyone to hold a digital asset.
Each capability is usable on its own and considerably more useful together — because they share one ledger, one identity graph, and one set of controls.
Models read every transaction as it happens — surfacing anomalies, forecasting cash position, and flagging exposure before it lands on a month-end report.
Accept and send money across every local rail through one API. Routing picks the cheapest successful path per transaction and retries intelligently when a rail degrades.
Put accounts, wallets, cards, and payouts inside your own product. Your users never leave your interface; the regulated machinery stays on our side of the line.
Underwrite from live cash-flow behaviour rather than collateral. Working-capital and receivables decisions return in seconds, with the reasoning attached for your credit committee.
A tamper-evident settlement ledger gives counterparties one shared record they can each verify cryptographically — proof for disputes and audit, with no digital-asset exposure.
Documented, versioned, and predictable. Typed SDKs, signed webhooks, sandbox parity with production, and errors that tell you what to do next.
Most financial infrastructure reaching this market was designed somewhere else and localized afterwards. That shows up in the edge cases, and edge cases are where money goes missing.
QRIS, BI-FAST, virtual accounts, and e-wallets are modelled natively in the core ledger — not mapped onto a foreign payment object that loses their semantics.
Because our models sit where the money moves, they see the whole picture — and act inside the authorization window instead of reporting on it the next morning.
Idempotency keys, exactly-once ledger writes, automatic reconciliation, and graceful degradation when an upstream rail is unavailable. The boring parts are the product.
In-country data residency, supervisory-ready audit trails, and licensed partner rails were designed in from the start rather than added under deadline.
Payments, lending, and embedded accounts write to the same system of record — so your balance, your risk view, and your books never disagree with each other.
Design partners get direct access to the engineers building their integration, a shared channel, and roadmap influence while we are still small enough for that to mean something.
Our design-partner cohort spans the sectors where transaction volume is high, margins are thin, and the finance function is still doing by hand what software should do by default.
Security is not a section in our documentation — it is the set of constraints the architecture was drawn against. Here is what that means concretely.
TLS 1.3 in transit, AES-256 at rest, and envelope encryption with regularly rotated keys held in a hardware security module.
Primary storage and replicas stay in-country, with processing and retention designed against Indonesia's personal data protection law.
Architected against Bank Indonesia and OJK expectations for payment and lending systems, operating through licensed partner institutions.
Every request, ledger entry, and model decision is recorded and replayable — so an auditor can reconstruct exactly what happened and why.
SSO with hardware-key MFA, narrowly scoped tokens, and break-glass procedures that are time-boxed and reviewed after every use.
Multi-zone deployment, continuous backup with tested restores, and degraded-mode operation so a single rail outage never stops settlement.
On certification: our SOC 2 Type II and ISO/IEC 27001 readiness programmes are underway. We will publish attestations when they are issued, and we will not claim them before then. Design partners can request our current security documentation and architecture review under NDA.
We are onboarding a limited number of design partners before public launch. If you are building something where money movement matters, we would like to hear about it.