Banks have a problem that is about to get a deadline. The cryptography protecting today’s financial systems will not survive the arrival of powerful quantum computers, and regulators have started putting dates on the fix. On November 16, 2026, Toronto-based Pelican AI Corp. offered its answer: Pelican Cortex, a new AI platform for financial crime prevention that is designed to stay auditable and verifiable through the post-quantum transition.
Cortex is not a chatbot or a generic AI assistant. It is built for one of the most demanding jobs in banking — deciding, payment by payment, whether money moving through the system is legitimate or criminal — and doing it in a way a regulator can check years later. Pelican says it intends to make Cortex the foundation of its whole product portfolio.
It is a launch worth understanding, because it sits at the intersection of two of the biggest stories in enterprise tech: AI agents making real decisions, and the long, expensive migration to quantum-safe cryptography. But it is also a launch with honest limits: Pelican disclosed no pricing, no deployment dates and no named customers. Here is what Cortex actually is, who it is for, and what to watch before calling it a success.
What Is Pelican Cortex?
Pelican Cortex is a sovereign, auditable and quantum-ready AI platform for financial crime prevention. In plain terms, it is software that watches payments as they move, flags the ones that look like fraud or money laundering, and keeps a record of every decision it makes — a record designed to survive the day today’s encryption stops being safe.
The word “quantum-ready” does the heavy lifting here. Today’s banks protect data with cryptography that a sufficiently powerful quantum computer could one day break. A decision record that cannot be verified ten years from now is useless to a regulator, so Cortex is designed to protect its evidence records with quantum-safe cryptography from the start.
The platform is built on a simple principle, as Pelican’s official announcement puts it: “the model proposes, the rules decide.” Proprietary AI models trained for financial crime read each payment and assess the risk. Rules — the bank’s own policies and regulatory requirements — make the final call. That split matters, because it keeps a human-designed control in charge of the decision, rather than letting a model act on its own.
Pelican AI is a Toronto-based company listed on the Canadian Securities Exchange (CSE: PEL). The company says its technology is trusted by major financial institutions in more than 55 countries and that it has over 30 years of proprietary AI expertise — a long track record in a field where most AI vendors are only a few years old.
Three Capabilities on One Platform

Cortex brings three capabilities together on a single platform:
Payment and sanctions screening. Every payment is checked against sanctions lists and watchlists before it moves. This is the bread-and-butter compliance work every bank must do, and doing it badly is how institutions end up with enormous fines.
Anomaly detection for fraud and money laundering. The platform looks for patterns that suggest fraud, money mule networks and money laundering — including on real-time payment rails, where money moves in seconds and there is no time for a human to review each transaction first. Real-time payments are the hardest place to catch crime, because the window to stop a bad payment is measured in moments.
Global regulatory compliance. The third capability is the least glamorous and arguably the most important: keeping the whole process aligned with the rules in each jurisdiction where a bank operates, and producing the evidence to prove it.
The practical promise is that a bank gets all three in one system instead of bolting together separate screening, fraud and compliance tools — each with its own data, its own alerts and its own audit trail. A single platform means a single version of the truth about any given payment.
That matters because the same week brought a reminder of how fast AI is spreading through enterprise software. ServiceNow’s Autonomous CRM Engine and SAP’s TechWolf acquisition both point the same way: AI agents are moving from demos into the systems where real work happens. Cortex applies that same shift to the most regulated corner of the enterprise — and in that corner, an agent that cannot explain itself is a liability, which is why the audit trail is the product, not an afterthought.
Why “Quantum-Ready” Matters Now
Two regulatory timetables are driving the timing of this launch, and both are worth knowing about.
In January 2026, the G7 Cyber Expert Group published a coordinated roadmap to post-quantum cryptography for banks, payment processors and their technology vendors. It points critical financial systems toward quantum-safe cryptography by 2030 to 2032, and the whole sector by 2035. That sounds far away, but migrating a bank’s cryptography is a multi-year project — institutions that wait until 2029 will be too late.
In Canada, OSFI Guideline E-23 on model risk, which covers AI and machine learning models, takes effect in May 2027 — less than six months from this announcement. E-23 will require Canadian financial institutions to show they understand, test and govern the AI models they use. An AI platform that cannot explain its decisions will struggle under that rule.
Cortex is designed for that timetable. Every decision is captured in a tamper-evident evidence record that Cortex is designed to protect with quantum-safe cryptography, so an examiner can verify it years later, even as cryptographic standards change. That is the feature Pelican is betting on: not just catching fraud today, but proving — years from now, under different cryptography — that the decision was sound when it was made.
The timing also fits a wider pattern. The same week saw Alibaba’s Qwen4-Max flagship model and NVIDIA’s Nemotron for Science — very different bets on the same shift toward AI doing real work. And as agents move into real workflows, securing AI agents in the enterprise is becoming just as important as deploying them — a point that applies directly to AI that decides whether a payment is criminal.
Sovereign by Design: Data Stays Where the Bank Is

Cortex is designed to run where a bank’s data already lives: on-premise, air-gapped or in a private cloud, with no dependency on outside AI model APIs. That is a deliberate choice, and it matters for the buyers Pelican is targeting.
Many banks are not allowed — or not willing — to send payment data to an outside AI provider’s servers. A sanctions screening decision involves some of the most sensitive data a bank holds: who paid whom, how much, and when. Running the AI inside the bank’s own walls, on its own infrastructure, removes that objection entirely.
The platform is designed to connect to Canadian payment rails — Lynx, ACSS/AFT, Interac and the Real-Time Rail — as well as international formats such as ISO 20022 and SWIFT. That coverage is the practical test of a financial crime platform: it has to see the payments where they actually move, not just in a demo environment.
Pelican intends to offer Cortex to its existing financial institution customers, which span more than 55 countries, as well as to new customers. The existing customer base is the real distribution advantage here — a new platform sold to banks that already run Pelican’s software has a much shorter road to deployment than one sold cold.
The Honest Limits: What’s Still Unproven
A launch announcement is not the same as a proven product, and this one leaves the most important questions unanswered.
No deployment dates. The release gives no date for when any institution will actually deploy Cortex. Pelican describes it as a platform it intends to make the foundation of its portfolio — language that suggests an early-stage product, not a deployed one generating revenue.
No financial terms. The announcement provides no financial terms or quantified revenue impact. There is no pricing, no indication of what a deployment costs, and no way to judge whether the offer is affordable for a mid-sized bank or only for the largest institutions.
No named customers. Pelican disclosed no customer names, so there is no way to tell who is actually using it. The 55-country customer figure refers to Pelican’s existing business, not to Cortex deployments.
No independent validation. The claims about quantum-safe evidence records and anomaly detection are Pelican’s own. No independent benchmarks or third-party evaluations were cited in the announcement, and the platform’s fraud-detection accuracy has not been publicly measured.
As Wisevoter’s report on the launch notes, the source gives no deployment timelines — and that is the gap that matters most. A platform can be well designed and still sit quiet if banks are slow to adopt it, or if its models do not match what compliance teams actually need.
What to Watch Next
Three signals will tell us whether Cortex is a real business or just a polished launch.
Named deployments. The first bank that publicly moves real payment screening onto Cortex will matter more than any press release. Watch for Canadian institutions first, given the OSFI E-23 deadline in May 2027 — and watch whether the deployment covers real-time rails, where the fraud problem is hardest.
Published pricing and packaging. Until Pelican says what Cortex costs, smaller institutions cannot judge whether it is genuinely affordable. Pricing is the difference between a platform anyone can try and a platform only large banks can afford.
Independent validation. Fraud-detection accuracy figures, audit results, or a regulator’s public acceptance of Cortex’s evidence records would let buyers compare it with rival systems. In financial crime prevention, a vendor’s own claims are the starting point, not the verdict.
Until those appear, the fair verdict is cautious optimism: a well-timed product aimed at a real regulatory deadline, from a vendor with a long track record in the space — with the proof of adoption still to come. More details are on Pelican’s Cortex information page.
Frequently Asked Questions
What is Pelican Cortex?
It is Pelican AI’s sovereign, auditable and quantum-ready AI platform for financial crime prevention, combining payment and sanctions screening, fraud and money laundering detection, and global regulatory compliance on one system.
What does quantum-ready mean here?
It means Cortex is designed to protect its tamper-evident decision records with quantum-safe cryptography, so an examiner can verify a decision years later even after today’s encryption standards are replaced.
What does “the model proposes, the rules decide” mean?
Pelican’s AI models assess the risk in each payment, but the bank’s own rules and policies make the final decision — keeping human-designed controls in charge rather than letting a model act alone.
Which payment systems does Cortex connect to?
It is designed to connect to Canadian rails including Lynx, ACSS/AFT, Interac and the Real-Time Rail, plus international formats such as ISO 20022 and SWIFT.
When does OSFI Guideline E-23 take effect?
May 2027. The guideline covers AI and machine learning model risk at Canadian financial institutions, and it is one of the regulatory deadlines driving demand for auditable AI platforms like Cortex.
How much does Pelican Cortex cost?
Pelican has not disclosed pricing. There is no public price list for the platform yet.
The Bottom Line
Pelican Cortex is a serious, well-timed bet on the future of financial crime prevention. It targets a real and dated problem — banks must move to quantum-safe cryptography by the early 2030s and satisfy AI model-risk rules by May 2027 in Canada — and it does so with a design that regulators should like: auditable decisions, human-set rules in charge, and data that never has to leave the bank.
The caveat is just as real. Without deployment dates, named customers, pricing or independent validation, nobody outside Pelican can yet judge whether the platform is adopted, affordable or accurate. The regulatory deadlines are real, the company’s track record is real, and the demand for auditable AI in banking is well established. The next chapter — who deploys it, and how well it works — is the one worth watching.

