Follow a private loan from borrower to investor and the glamorous language disappears quickly. What remains is a small blizzard of files: loan tapes, payment histories, eligibility rules, covenants, cash-flow models and reports. The assets may be bespoke. The workflow is often familiar. A spreadsheet arrives by email, someone checks its columns, someone else reconciles the totals, and an investment professional waits for a clean enough view to make a decision.
Altin Kadareja built his company in that waiting time. Before Cardo AI, he spent nearly a decade moving through the machinery of European finance. He worked in product innovation at Allianz Bank Financial Advisors, analysis at Intesa Sanpaolo, risk and consulting at Prometeia, and asset management at BlackRock. Different desks gave him different angles on the same awkward truth: highly trained people were spending too much of the day managing information instead of interpreting it.
The irritation was useful. In 2018, Kadareja co-founded Cardo AI to give asset-based finance and private credit a purpose-built operating layer. The company collects data from disparate sources, validates it, applies transaction rules, models cash flows and keeps portfolios visible after a deal closes. The pitch is operational before it is futuristic. Give every number a path back to its source. Make routine tests repeatable. Let specialists use their hours on judgment.
An apprenticeship in the seams
Kadareja is from Albania, then studied in Milan at Bocconi University. He completed an undergraduate degree in business administration and a master's in economics and management of innovation and technology. A semester at Copenhagen Business School added another country to the itinerary. In 2017, after years in industry, he completed an executive program in risk management at Imperial College Business School in London.
That mix explains the company better than a generic founder origin story would. His education joined management, economics and technology before those fields became a job description. His career then placed him at the seams between products, risk, data and investment decisions. Cardo AI did not begin with a broad ambition to “do AI for finance.” It began with a workflow he knew closely enough to find annoying.
A city plan for financial data
The name Cardo comes from ancient urban planning. In a Roman city, the cardo was the north-south axis, crossing the decumanus and organizing movement around a central route. Kadareja's company borrowed the term as a product metaphor. Private credit contains many participants and many handoffs: originators, lenders, asset managers, servicers, trustees and fund administrators. A shared data layer can act as the axis connecting them.
The metaphor also reveals a preference for infrastructure over spectacle. A city street is valuable because other activity can happen on it. Cardo AI's software sits beneath borrowing-base calculations, covenant monitoring, portfolio analysis and investor reporting. Each function sounds narrow. Together they determine how quickly a credit team can understand what it owns and whether something has changed.
“Our mission is to make private markets more transparent and efficient through ‘intelligent’ technology.”Altin Kadareja, 2026
Transparency is an unusually concrete word here. It means the analyst can inspect the input, the risk team can see the rule, the operator can repeat the process, and the auditor can follow the result backward. In public markets, prices and standardized disclosures do much of that organizing work. Private credit is private partly because each deal carries its own terms, documents and rhythms. Flexibility is a feature, but operational fragmentation is the bill.
The American chapter
Cardo AI expanded into the United States in 2024. That November, it announced a $15 million Series A co-led by Blackstone Innovations Investments, FINTOP Capital and JAM FINTOP, with participation from Andy Horwitz and Kevin MacDonald. Blackstone also became an enterprise client, using Cardo AI technology in its Credit & Insurance operations for direct lending and asset-based finance transactions.
The investor mix was part capital and part distribution map. FINTOP and JAM FINTOP brought a network of regional and community banks. Blackstone brought the perspective of a large alternative-credit operator. Kadareja had moved to New York, placing the founder closer to a market where private credit was growing and where institutions were looking for ways to manage more varied assets without adding the same amount of manual work.
By 2025, Cardo AI had joined that New York lab, a 12-week program run by Accenture and the Partnership Fund for New York City. The company also collected industry awards for debt portfolio management technology and service provision. Awards are snapshots. More revealing is the company's team design: credit specialists sit beside software, data, infrastructure and machine-learning practitioners. Domain experts define the logic. Engineers make the logic repeatable.
The transferable operating lesson: put the practitioner beside the builder. The exception one person remembers can become the rule everyone can test.
AI, with receipts
Kadareja's public argument for AI is careful about trust. A model may clean a feed, detect an anomaly or extract a term from an unstructured document. But a financial institution still has to explain the result. His compact version is that auditors and regulators will not accept “the model said so.” The useful system moves faster while keeping the source, rule, check and audit trail visible.
That constraint makes the work harder and more interesting. Consumer AI can tolerate an answer that feels plausible. Credit cannot. A decimal in the wrong column can change a borrowing base; an old rating can hide a changing risk; a covenant sitting in prose can be missed until it matters. Cardo AI's bet is that artificial intelligence belongs inside the workflow, bounded by traceability rather than floating above it as a clever interface.
In 2026, Kadareja described the next shift as a move from static data management to continuous, decision-grade data. Reporting after the fact is not enough when collateral pools and borrower performance keep changing. Information needs to be structured and checked frequently enough to support valuation, risk and portfolio choices while those choices are still available.
The practical consequence is a different tempo for credit work. Instead of assembling a retrospective picture at the end of a month or quarter, a team can keep eligibility, concentration and performance measures in view as new information arrives. That does not remove disagreement about a loan or a portfolio. It improves the object around which people disagree. Kadareja's version of modernization is therefore as much organizational as technical: operations, risk and investment teams work from the same evidence, and fewer decisions begin with a debate over which spreadsheet is current.
“The market needs to be able to scale faster and not be limited by what the deal team could produce.”Altin Kadareja on The Credit Clubhouse, 2026
A founder still reading the conditions
There is a lighter detail in Kadareja's public biography: when he needs new ideas, he travels in search of the next good surfing wave. It is tempting to force the metaphor, so only a small one is necessary. Surfing rewards attention to conditions. So does lending. Neither improves because someone insists the environment should match yesterday's model.
His international route has demanded a similar flexibility. Albania supplied the starting point and later became part of the company's technology base. Milan supplied education and the first long chapter of finance work. London added training and a company office. New York became the center of the American push. Cardo AI now describes a team spread across five countries, with credit expertise and engineering talent deliberately mixed.
In a 2026 manifesto, Kadareja opened with a memory: “Seven years ago, people told me I was crazy!” The disputed premise was that artificial intelligence could work in structured finance, a market full of bespoke deals and cautious institutions. Seven years later, his case is less about whether AI can produce an answer. It is about whether the data underneath the answer is clean, the workflow around it is usable, and the reasoning can be inspected.
That distinction gives the story its shape. Kadareja did not leave finance to escape its details. He built deeper into them. Cardo AI treats the boring parts as consequential: column names, document terms, validation rules, reconciliations, reporting cycles. Those are small things until billions of dollars depend on them.
The ambition is larger than eliminating a few spreadsheets, but it remains grounded in the people opening them. If private credit continues to expand, teams will need to process more assets, more structures and more reporting without turning every new strategy into another manual exception. Kadareja wants the operating layer to absorb that complexity. The expert should still make the decision. The system should make sure the expert can see what the decision is made of.