Every doubt, answered

The category made one foundational error and then compounded it. It chose to manufacture trust between strangers, which forces you to buy both sides of every transaction, forever, at retail prices, on a product most customers use once. Adverse selection follows from that. Ruinous acquisition costs follow from it. Fragile retail capital follows from it. The absence of any real network effect follows from it. So does the regulatory exposure that eventually killed the survivors.

Pari is not a better marketplace. It is an inversion on the two axes that mattered. We do not create the relationship, we instrument one that already exists, so there is no trust to originate and no stranger risk to price. We do not buy customers, we enter through an institution that already owns the relationship, so there is no second acquisition bill and no two-sided balancing problem.

The honest way to test that claim is one failure mode at a time rather than in the abstract. We published all nine, with what each platform did and what we do instead, on the graveyard page.

Full argument and sources: Learning from the Graveyard

CircleLending is the most important company in our analysis, because it was right about the thing everyone else was wrong about and it still did not survive. Founded in 2000, it formalised and serviced loans between friends and family, and explicitly rejected the stranger-lending model that Prosper, LendingClub and Zopa would later build. Virgin took a majority stake in 2007. It was dissolved in November 2010 without a formal announcement.

It failed on channel, not on product. It sold direct to consumers, and a family loan is not a product people wake up wanting: it surfaces during a life event, in a conversation with someone the family already trusts. It positioned as a documentation utility, a one-time, low-price transaction with no recurring relationship. And it had no institutional partner whose economics improved when it succeeded. Nobody else in the value chain made money, retained a client, or reduced a risk when a family loan got papered properly.

CircleLending proved the primitive and disproved the channel. We took the primitive and replaced the channel. The advisor is already sitting with the family when the question comes up, already knows the balance sheet, has a fiduciary reason to care whether the loan is structured properly, and has a commercial reason of their own to want the next generation in the room.

Full argument and sources: Learning from the Graveyard

It was founded by a former CircleLending executive to carry the family-lending model forward in the narrower domain of intra-family real estate loans. That is the same channel decision made a second time, on a smaller surface: consumer-direct, one asset class, documentation and servicing sold as a transaction rather than embedded in a workflow somebody else already runs.

We read it as two things at once. It is evidence that the underlying behaviour is durable enough to support a business for well over a decade, which matters. And it is evidence that documentation alone does not compound, which is the lesson we took from it. The difference we are betting on is the full stack (underwrite, document, service, report) delivered through an institution that has its own reason to want it to work.

Vouch built its company on a substitute for knowing the borrower: friends and family sponsored an applicant and put money at risk, which was supposed to improve repayment. It was intellectually elegant and it did not produce volume.

We read that as support rather than a warning. Every attempt to find a cheap proxy for actually knowing the borrower, whether alternative data, social graphs, alma mater or employment history, was trying to approximate information that in a family is simply present. The lender has decades of unmediated knowledge of the borrower's income, reliability and circumstances. That is not a proxy for underwriting. It is better information than any institution can buy.

Several people tried, and each did it consumer-direct. Beyond the channel error, three conditions had to be true at the same time and only recently were: banking infrastructure mature enough to give a small team programmatic payment rails, compliance tooling cheap enough that an early-stage company can credibly run a documented BSA/AML program and pursue SOC 2, and an advisory channel large enough and under enough succession pressure to want the product.

There is also a less flattering answer that we think is true. The segment is small by headcount and unglamorous, and the last two decades of consumer fintech rewarded the opposite instinct.

Full argument and sources: Sharing Economy 3.0

Deliberately. The peer-to-peer generation went after the largest visible population of underserved borrowers. We went after the smallest population with the largest amount of capital moving between them.

Headcount is the wrong numerator. What matters is capital volume per acquired relationship, multiplied by frequency, minus loss, minus the cost of acquiring the relationship at all. A platform paying roughly $400 to acquire a borrower who takes one $8,000 loan, with high loss rates and no natural repeat, only works at extreme scale and only while capital markets cooperate. A platform acquiring one advisory firm, and with it several hundred to several thousand high-net-worth households, for a fixed integration cost is a different business that looks similar on a slide.

Compliance, servicing, collections, tax reporting and support cost roughly the same per loan whether the loan is $4,000 or $400,000. The category chose the $4,000 loan. Intra-family loans in an advised context are large and long-dated: property, business capitalisation, estate structures, education.

Full argument and sources: Learning from the Graveyard

The transaction recurs across a family lifecycle rather than annually: a first home, a business, a tax event, an equalisation between siblings, a second property. But we would not build a revenue model on household frequency, and we do not.

The unit that recurs is the firm. An advisory firm has a continuous flow of these events across its client base, and the servicing of an existing loan is ongoing rather than one-time. A loan booked today is serviced, reported and tax-tracked for years.

This is the sharpest commercial objection and it deserves the real answer rather than a slogan. In the short run, capital lent is capital not managed.

The framing also contains an error worth correcting first. The capital does not leave the advisor's field of view. A loan structured through Pari stays inside the firm's ecosystem: it is originated in the advisor's workflow, it appears on the client's financial profile, and it is reported in the advisor's dashboard as a performing note with a rate, a schedule, a balance and a counterparty. What actually leaves an advisor's view is the informal version, the wire that goes out with no document behind it and simply disappears from the household balance sheet. Structuring the loan converts an invisible outflow into a tracked household asset.

And at many firms that asset is billable. A large share of advisory firms charge on assets under advisement alongside assets under management, precisely so they can be paid for guidance on holdings they do not custody: real estate, private company stock, held-away accounts, concentrated positions. A documented, serviced, valued intra-family note is the kind of asset those schedules exist for. An undocumented family transfer is not. Whether a particular note falls under a particular firm's AUA schedule is a question for that firm's own fee agreement and its compliance team, but the structural point stands: structuring the loan can move capital out of the untracked column and into the advised one, rather than out of the practice.

What it buys beyond that is the moment advisors are most afraid of. Retention of a living client runs near 97 percent a year. Retention through the generational handoff collapses. Cerulli finds more than 70 percent of heirs are likely to change advisors after inheriting, and that only 27 percent of future beneficiaries plan to keep their benefactor's advisor, falling to 20 percent among those who have already inherited. The stated reason is not performance: only 10 percent cite unmet investment needs, while half already had their own advisor and 28 percent had no relationship with the benefactor's advisor at all.

Advisors know this. Natixis found in April 2026 that 41 percent of US advisors regard the wealth transfer as an existential threat to their practice, and 22 percent say they have already lost significant assets to generational attrition. The same research locates the leak precisely: assets are retained around 78 percent of the time when a spouse inherits, because the spouse was in the room. It breaks going down a generation, for the same stated reason every time.

An intra-family loan is one of the very few financial events that requires the next generation to be in the room, with the advisor, on a matter of real consequence to them, as a principal rather than a spectator. Not an annual review they skip. Not an estate document they never read.

We would not ask anyone to accept that on argument alone. It is an empirical question and the relevant evidence is our conversion from signed intent to live implementation, and advisor retention data over time. It should be evaluated on results.

Full argument and sources: Learning from the Graveyard

Two different questions get collapsed into one acronym, and separating them is the whole answer. Distribution answers how a family finds Pari. Monetization answers who pays. For Pari those have different answers.

Distribution is B2B. Pari is free for private wealth firms, permanently. No firm is ever invoiced, no firm resells the product, and no firm takes a margin. The advisory firm is an introduction channel, and its incentive to make the introduction is its own retention problem rather than a revenue share.

Monetization is B2C. The family pays Pari directly, through transaction fees on structuring and servicing a loan, and through subscriptions. Pari sets its own pricing and collects it from the end customer.

This is deliberately not B2B2C, and the distinction is not cosmetic. In a B2B2C model the business partner owns the end customer, resells the product and takes a cut. That stacks margin, makes the vendor dependent on a partner's sales motion, and leaves the vendor without a direct relationship with the person actually using the product. Pari has a direct relationship with the family, prices to the family, and is paid by the family. The firm sits alongside that relationship rather than on top of it.

Five co-founders. The founding team is 100% first-generation American, immigrant or international, led by a US Army veteran founder, with two software engineers carrying both consumer and enterprise experience from Accenture, IBM, Snapchat and JP Morgan, an L5 product manager from Google, and a systems and operations MBA from JP Morgan.

Yoshi Mua, Founder and Chief Executive Officer. Madson Cardoso, Co-Founder and Chief Technology Officer. Yannick Wade, Co-Founder and Chief Product Officer. Megh Patel, MBA, Co-Founder and Chief Growth Officer. Hisham Mohammed, Co-Founder and Senior Software Engineer.

The relevance to this particular business is the mix. Intra-family lending needs consumer product instincts, enterprise-grade engineering, a compliance posture that a fiduciary firm will pass through vendor diligence, and the operational discipline to service loans correctly for years. Those are four different disciplines and the team was assembled against them.

It is, which is why the roadmap is sequenced rather than parallel: Lend, then Ledger, then Account, then Debit, then Credit.

Lending comes first because it is the only entry point that is simultaneously already happening at enormous volume, high-stakes enough to justify adopting software, legally structured but under-served, and generative of the data every later stage requires. A family bank that has serviced eighteen months of intra-family loans has an underwriting history, a payment history, a governance record and a compliance file. Those are the raw materials for everything after it.

Nothing about the sequence can run in reverse. A debit card in a family's name with no lending history underneath it is a novelty product. The later stages carry regulatory dependencies we name rather than hide.

Full argument and sources: The Decentralization Error

They are open marketplaces aimed at subprime and underbanked borrowers, typically for very small sums, often under $500. That is the same obvious white space the last generation ran at, and it is empty for the same reason it was empty then: those borrowers are expensive to acquire, hard to underwrite and adversely selected, and the loan sizes cannot carry the compliance and servicing stack built around them.

We are not competing with them for the same customer. They are at the maximum-headcount, minimum-capital-density end of the market. We went to the opposite end deliberately, which is the F4 and F6 argument on the graveyard page.

They do, and the document is the cheapest component of the problem. This is precisely the trap the best-capitalised predecessor fell into: papering and servicing a loan is a real service, but it is a one-time, low-price, low-frequency transaction with no recurring relationship, which is why the successor products narrowed to documentation in a single asset class.

What a family actually needs after the note is signed is the part nobody sells them: an amortisation schedule that stays correct, payments that happen without a parent chasing their child, an interest rate that stays compliant with Section 7872, a tax position that survives an audit, and a defined path for modification, forbearance or forgiveness when circumstances change.

If it were a document and a payment, nothing would. The defensible surface is the rest of it: affordability underwriting on real-time data, servicing, tax reporting, the compliance program, the bank relationships, the advisor integration, and the loan performance data that accrues from all of it over time.

The structural answer is the one we care about more. We operate inside two pre-existing networks with genuinely high exit costs. Families do not churn out of families, and clients do not churn out of advisors. The moat is not the software.

No, and the reason is structural rather than argumentative. Pari does not raise, pool, hold or intermediate investor capital. The lender's capital is the lender's own, already earmarked for a family member. There is no yield promise, no note issued to the public and no investor to protect.

That is the surface which produced the SEC's 2008 order against Prosper, and we do not have it. The capital was never ours to lose, which removes the failure mode rather than mitigating it.

Full argument and sources: Learning from the Graveyard

Both boundaries are real and we stay on the correct side of them deliberately. Loan documents are attorney-drafted templates that a family completes for itself, which is the long-settled self-help posture rather than individualised legal advice. The family's own attorney and CPA remain the advisors of record, and in the advised channel one or both are usually already involved.

Rate and threshold calculations are informational computation against published IRS figures. We track the Applicable Federal Rate and the annual exclusion because they change and because getting them wrong is what exposes a family on audit, not because we are forming a view on anyone's tax position. Every tax figure we publish is dated, sourced and carries a not-advice disclaimer.

An individual lending to their own child is not a creditor who regularly extends credit, so the Truth in Lending Act and Regulation Z generally do not attach to the loan itself. Charging at or above the Applicable Federal Rate places family loans far below any state usury ceiling, because the AFR tracks Treasury yields rather than consumer credit pricing.

The obligation that does attach arrives later in our roadmap. Furnishing repayment history to credit bureaus would make Pari a furnisher under the Fair Credit Reporting Act, with real accuracy and dispute-handling duties. We would rather name that now as a build requirement than discover it as a surprise, and it is one reason the roadmap is sequenced the way it is.

The lender never sees the borrower's financial data. This is the single most important design decision in the product and it is not a setting, it is the architecture.

Affordability underwriting returns an answer about whether a repayment schedule fits, not a disclosure of what anyone spends money on. Early-warning nudges are private and borrower-facing: if a borrower's cash flow is tightening, the borrower hears about it, not the parent. A parent learns whether the loan is affordable and whether payments are being made, which is what a lender is entitled to know, and nothing else.

The closest available evidence is Mission Asset Fund's formalised lending circles, which report roughly 99 percent repayment and an average 168-point credit-score improvement among participants. That work is drawn from low-income immigrant communities with no capital and no credit files, so the mechanism producing the result was documentation and reporting rather than wealth.

Set against informal family lending, where Bankrate found 44 percent of lenders lost money outright and 26 percent had a relationship damaged, the gap is large and it is attributable to structure.

We also state the inverse as a falsification test. If documented, automated family loans turn out to default at rates comparable to the 40 to 50 percent range reported for informal arrangements, the core value proposition is disproved. That number is measurable within the first several hundred funded loans and our instrumentation is designed to produce it.

Full argument and sources: The Decentralization Error

It is the oldest form of credit there is, and the only thing missing is infrastructure. The National Association of Realtors found 22 percent of first-time buyers funded part of a down payment through a gift or loan from a relative or friend in 2025, while the median first-time buyer age reached a record 40 and the first-time buyer share fell to a record low 21 percent.

At the high end the transaction is more technical and higher stakes. Below-market family loans have been governed by Internal Revenue Code Section 7872 since 1984. Charge below the Applicable Federal Rate and the shortfall is imputed and treated as a gift. Fail to document and service the loan properly and the whole structure is exposed on audit.

So the household faces a transaction that is common, large, legally technical, tax-sensitive, emotionally loaded, and currently handled with a template note and a spreadsheet. That is the definition of a segment where infrastructure creates value.

Full argument and sources: The Bank of Mom & Dad

In the advised segment the driver is usually mechanical rather than emotional. A loan is documented because Section 7872 requires an AFR-compliant rate to avoid an imputed gift, because the estate plan depends on the note existing, or because siblings need to see that an advance was an advance. The presence of an advisor is itself evidence of a planning motive rather than a suspicion motive.

The broader point is that structure is what makes generosity safe rather than what makes it conditional. A parent who wants the option to forgive a loan later is better served by a loan with a defined forgiveness path than by an ambiguous transfer nobody can characterise afterwards.

The premise is right and the conclusion is backwards. Family lending damages relationships today because it is unstructured: no schedule, no documentation, no defined delinquency process, and a lender who has to personally chase repayment from someone they love.

Every one of those failure modes is an absence of infrastructure. An amortisation schedule removes ambiguity about what is owed. Automated servicing removes the parent from the collections role. A written agreement converts an emotional negotiation into a policy decided in advance, while everyone is calm. A defined forgiveness path makes generosity deliberate rather than a slow surrender.

The evidence points the same way. Socially embedded lending performs extremely well when it is structured. It performs badly when it is a handshake.

Full argument and sources: The Decentralization Error

It is the right question to ask about any product that moves money from an older person to a younger one, and the answer is the part of our model that is easiest to overlook: there is a fiduciary in the transaction.

Pari is distributed through registered investment advisors, who are legally obligated to act in the client's interest. The lending parent is the advisor's client. An advisor who sees a proposed loan that does not fit the parent's own liquidity, retirement needs or estate plan has both the information to notice and a legal duty to say so. That is a materially stronger control than anything a consumer-direct product can offer, and it is a control that the informal alternative has none of. Elder financial abuse overwhelmingly happens in the absence of a professional, not in front of one.

The product reinforces it rather than relying on it alone. Affordability underwriting runs on the lender as well as the borrower, so a loan the parent cannot actually afford to make is visible before it is made rather than discovered afterwards. Both parties execute independently, so the arrangement cannot be assembled by one side. And every loan produces a dated, documented record of what was agreed, by whom, and on whose advice.

The comparison that matters is the counterfactual. An undocumented cash transfer made under pressure leaves no trace, no record of consent, no affordability check and nothing for a family member, an advisor or an investigator to find later. A documented loan, underwritten for the lender and witnessed by a fiduciary, is the harder thing to abuse, not the easier one.