Learning from the Graveyard

This is a Pari white paper. It is published in full, with its sources, because the argument only counts if the working is visible. Nothing here is investment, legal or tax advice.

Peer-to-peer lending was one of the most heavily funded consumer fintech theses of the last twenty years. It is now, as a category, essentially extinct in every major market. The interesting question is not whether it failed. It is why the same failure kept happening across different countries, different regulators, different credit cycles, and different founding teams.

Our answer is that the category made one foundational error and then compounded it seven more times. The foundational error was choosing to manufacture trust between strangers, because doing so requires you to buy both sides of every transaction, forever, at retail prices, on a product that most customers use once. Everything else follows from that. Adverse selection follows from it. Ruinous acquisition costs follow from it. Fragile retail capital follows from it. The absence of any real network effect follows from it. And the regulatory exposure that eventually killed the survivors follows from it too.

Pari is not a better peer-to-peer marketplace. It is a deliberate inversion of the model on the two axes that mattered most.

We do not create the relationship; we instrument one that already exists. Intra-family lending is not a market we have to manufacture. Families have lent to each other for as long as there have been families, and they do it today at a scale nobody measures, almost entirely undocumented. There is no trust to originate, no stranger risk to price, and no adverse selection problem of the kind that destroyed the open marketplaces. The lender already knows the borrower's income, their character, their history, and where they will be at Thanksgiving.

We do not buy customers; we enter through an institution that already owns the relationship. Registered Investment Advisors sit on top of exactly the households that do this kind of lending, retain those households at roughly 97% a year¹, and are structurally terrified of the one moment when that retention breaks: the intergenerational transfer².

That yields a defensibility structure the peer-to-peer platforms never had: two stacked networks, each with a genuinely high exit cost. Families do not churn out of families. Clients do not churn out of advisors. The moat is not the software. The moat is that we are inside two networks that people find very hard to leave.

Every failed P2P platform had to buy a stranger, price a stranger, and then buy another stranger. We are handed a family by an advisor who needs us to succeed.

---

A mortality table for peer-to-peer lending

Read the dates, not the logos. Almost none of these ventures were killed by a competitor. They were killed by the model: by capital that would not stay, by borrowers who cost more to find than they were worth, by regulators who correctly identified what the platforms actually were, and, in the two cases most relevant to us, by trying to sell a relationship product directly to consumers who were not looking for one.

Pari is not on this table. This paper exists so that it never is.

---

What actually happened, 2005-2021

Act one: the open auction, and what strangers do to each other

The original promise was disintermediation. Prosper opened to the public in 2006 with an auction: a borrower posted an amount and a reserve rate, individual lenders bid the rate down in $25 slices, and the market cleared. LendingClub launched into the same thesis. Zopa had already opened in the UK the year before. The pitch was that a crowd could price credit better than a credit officer.

It could not. Academic work on Prosper's early years found the platform's listing pool deteriorated on observable characteristics over time even as lenders got better at selecting within it, the textbook signature of adverse selection in an open market³. The most rigorous study of the 2008 cohort found something even more damning about the supply side: the mean internal rate of return on a Prosper lender's portfolio was approximately negative 4.1%, and the median lender had funded just six loans totalling around $350⁴.

Read that twice, because it contains the whole story. The typical retail lender lost money and never committed meaningful capital in the first place. A marketplace where the supply side is unprofitable and uncommitted is not a marketplace. It is a subsidy waiting to be withdrawn.

Then the regulator arrived. In November 2008 the SEC found that the notes Prosper had been issuing since January 2006 were securities sold without an effective registration statement, and entered a cease-and-desist order⁵. Prosper had already stopped accepting lender funding in October 2008 and restructured its entire operating model to resolve the uncertainty⁵. State regulators followed. The auction died with it.

Act two: the quiet institutionalisation

What replaced peer-to-peer was not peer-to-peer. It was a loan origination business with a retail-investor marketing department attached. Institutional capital (hedge funds, banks, asset managers) became the real funding source, and the retail "peer" became a rounding error. By June 2020, retail investors funded roughly 17% of LendingClub's loans, and that was a four-year high⁶.

The economics of the replacement model were brutal in a different way. Because the platforms now had to buy borrowers directly, at scale, in a market where every competitor was buying the same borrowers, acquisition became the dominant line item. LendingClub's sales and marketing expense ran to $268.5 million in 2018 alone, or 2.47% of loan originations⁷. SoFi spent $170 million on marketing in 2017, roughly $756 per acquired customer, mailing 15 million pieces of direct mail a month; industry estimates put LendingClub and Prosper in the $350-$450 range per customer⁸.

These are not the numbers of a network. They are the numbers of a paid-media arbitrage that works only while the arbitrage lasts.

Act three: 2016, when the capital stopped

The break came in 2016. LendingClub's founder-CEO departed amid an internal investigation. Prosper, valued at $1.9 billion the previous year, never profitable, with losses growing from $3 million in 2014 to $26 million in 2015, cut 28% of its staff, 171 people, and pulled its volume guidance⁹. Vouch, the most interesting of the relationship-underwriting startups, wound down after its investors declined to continue funding it¹⁰. The industry has been moving away from peer-to-peer since 2016, as one early market participant put it at the time of LendingClub's final retail exit¹¹.

Act four: the exits

The rest is a wind-down. The UK regulator introduced new rules restricting how P2P could be marketed to retail investors in 2019; most British platforms closed thereafter¹². Metro Bank acquired RateSetter in 2020 and closed retail lender accounts in April 2021¹³. Funding Circle paused retail lending in April 2020 and never brought it back¹³. LendingClub retired the Notes platform on 31 December 2020, telling investors plainly that under a prospective banking framework it was not economically practical to continue⁶. In December 2021 Zopa, the company that started the category, announced it was closing its P2P business, citing negative investor sentiment and tighter regulation squeezing margins¹⁴.

The control experiment: China

China ran the same model with almost no regulation and 6,000-plus platforms. It is the cleanest available test of what happens when you scale stranger-to-stranger credit without institutional constraint. Active platforms fell from 3,464 in 2015 to 343 by 2019 before the sector was effectively closed by the regulator¹⁵; roughly 15% of the more than 6,000 platforms ever established survived¹⁶. Ezubao alone took approximately $7.6 billion from around 900,000 investors before being exposed as a Ponzi scheme¹⁷.

Four regulators, three continents, one outcome. When the same model dies in every environment it is tried in, the environment is not the variable.

---

Eight failure modes, and what we do instead

Each of these is a distinct, independently fatal design decision. Most of the failed platforms made all eight. We consider a business defensible only if it is structurally immune to each one, not merely aware of it.

F1 · Trust: They tried to manufacture trust between strangers, then price it

An open marketplace has to solve the hardest problem in credit (asymmetric information) using only the signals a stranger will volunteer. Early Prosper listings combined a loan application with a personal narrative, and the crowd priced sympathy alongside credit grade. The observable quality of the listing pool worsened over time³, borrowers who could not get bank credit self-selected in, and the lenders bidding against each other had no mechanism to detect it.

Every subsequent attempt to fix this, whether alternative data, social graphs, alma mater or LinkedIn, was an attempt to find a cheap substitute for actually knowing the borrower. Vouch built its entire company on the substitute: friends and family sponsored a borrower and put money at risk, which was supposed to improve repayment¹⁸. It was intellectually elegant and it did not produce volume.

What Pari does instead. We do not price stranger risk, because there is no stranger. In an intra-family loan the lender has decades of unmediated information about the borrower: income, reliability, life circumstances, family obligation. That is not a proxy for underwriting. It is better information than any lender in the market has.

Our job is not to assess whether the parties should transact. They have already decided that. Our job is to make the transaction correctly structured, AFR-compliant, documented, serviced, and tax-defensible.

F2 · Cold start: They had to buy both sides of the market, forever

A two-sided marketplace between strangers has two acquisition bills and no natural mechanism for either side to bring the other. Retail lenders had to be recruited, educated, converted, and retained. Borrowers had to be bought outright through direct mail and paid media. Neither side referred the other, because there was no relationship between them to refer through.

Worse, the two sides had to be balanced continuously. Excess borrower demand meant unfunded listings and a broken product experience. Excess lender capital meant compressed yields and lender attrition. The platforms spent enormous sums keeping a see-saw level.

What Pari does instead. Both sides of a Pari loan arrive together, pre-matched, at the moment the family decides to act. The lender and the borrower are already in the same room, often literally. There is no matching problem, no balance problem, and no second acquisition bill.

This is the single largest structural difference between us and every platform in the mortality table above, and it is the reason our cost structure is not a function of paid media.

F3 · Capital: Retail investors were the wrong balance sheet

Retail P2P capital turned out to be the most fragile funding source in consumer credit. It was small, sentiment-driven, and quick to leave; the median Prosper lender committed around $350⁴. When the pandemic hit, investors cashed out and platforms shuttered; Zopa's chief executive named damaged sector reputation and negative investor sentiment as central to the decision to close¹⁴. One observer's summary of the end state in Britain was that there were plenty of borrowers and no investors¹².

The platforms also inherited the regulatory weight of that fragility. Retail-facing credit investment products attract exactly the supervision they attracted.

What Pari does instead. 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, and in many cases already sitting inside an advised portfolio. There is no yield promise, no note issued to the public, and therefore no securities-registration surface of the kind that produced the SEC's 2008 order against Prosper.

The capital was never ours to lose. That removes the failure mode entirely rather than mitigating it.

F4 · Segment: They optimised for user count in the least capital-dense segment available

The subprime and near-prime consumer looked like the ideal beachhead: enormous, visibly underserved, morally sympathetic, and easy to describe to investors. It was also the worst possible place to build a capital business. Small loan sizes, high loss rates, low repeat frequency, thin margins, aggressive competition, and, as LendUp discovered, the closest regulatory scrutiny in consumer finance.

LendUp raised money from GV, Andreessen Horowitz, Kleiner Perkins, PayPal and QED, and had its lending operations shut down entirely by the CFPB in December 2021 after repeated enforcement actions¹⁹. The quality of the cap table did not protect the model.

What Pari does instead. We went to the smaller, denser, non-obvious segment: high-net-worth families making structured intra-family loans, reached through their advisors. Fewer households, dramatically more capital per household, materially lower credit risk, and a transaction that recurs across a family's lifecycle: a home, a business, a tax event, an equalisation between siblings.

F5 · Acquisition: They paid subscription-business acquisition costs for a transactional product

A consolidation loan is close to a single-purchase product. The customer takes it, repays it over three to five years, and has no structural reason to return. Against that, the category was paying $350 to $756 to acquire each customer through direct mail and paid media⁸, in an auction where every competitor bid on the same keywords and the same mail lists.

The US Treasury's own review of the sector noted the obvious fix: white-label and co-branded partnerships with institutions can materially reduce customer acquisition costs for online marketplace lenders²⁰. Upstart eventually built its business on precisely that insight, partnering with banks and credit unions rather than competing with them for the same borrower²¹. Most of the category learned it too late.

What Pari does instead. Our acquisition unit is not a household. It is an advisory firm. One signed firm exposes us to its entire client base, and each subsequent family inside that firm carries near-zero incremental acquisition cost.

Our current signed letters of intent represent six firms, approximately $10.6 billion in AUM and roughly 4,700 high-net-worth client households, with 50-plus firms in active pipeline. That is six acquisition events for a four-figure household count.

F6 · Loan size: Small loans cannot carry the cost of the infrastructure around them

Compliance, servicing, collections, tax reporting, dispute handling and support cost roughly the same per loan whether the loan is $4,000 or $400,000. The category chose the $4,000 loan. That forced high rates to cover fixed costs, which selected for worse credit, which raised losses, which forced higher rates.

China's collapse was the terminal version of this dynamic at scale, with widespread fraud filling the gap between what micro-loan economics could support and what platforms promised investors¹⁶.

What Pari does instead. Intra-family loans in an advised HNW context are large and long-dated: property purchases, business capitalisation, estate-planning structures, education. The same compliance and servicing stack that would be uneconomic across thousands of micro-loans is comfortably carried by loans an order of magnitude larger, with materially better credit characteristics.

F7 · Regulation: They faced the regulator as principal, not as infrastructure

Peer-to-peer platforms occupied the most exposed possible position: issuing securities to retail investors on one side and extending consumer credit on the other, frequently through rent-a-charter arrangements that regulators and courts scrutinised heavily. Prosper drew an SEC cease-and-desist in 2008⁵ and later settled charges relating to overstated annualised net returns²². LendUp drew three separate CFPB actions and was ultimately shut down¹⁹. The UK regulator's 2019 rules effectively ended retail P2P in Britain¹².

What Pari does instead. Pari is infrastructure, not principal. We do not lend, do not issue securities, do not take deposits, and do not sell an investment product. We document, structure, service and report on loans between two consenting family members, in partnership with a chartered institution for movement of funds.

We have nonetheless built to institutional standards ahead of scale: a full BSA/AML compliance program authored by our General Counsel and BSA Officer, mapped into nine engineering workstreams covering CIP, beneficial ownership, CDD, OFAC screening, transaction monitoring, SAR workflow, recordkeeping, 314(a)/(b) response and governance, plus a SOC 2 audit in progress with an external auditor and a continuous-compliance platform.

F8 · Defensibility: They built funnels and called them networks

This is the deepest error and the one most often disguised. A peer-to-peer marketplace does get marginally better with scale: more capital means faster funding, more borrowers means more selection. But the individual user experiences almost none of that. Nobody chose LendingClub because their friends were on it. Nobody stayed because leaving was hard. There was no exit cost of any kind.

Network effects are the strongest of the four durable defensibilities available to technology businesses, and the research consensus attributes a large majority of technology value creation since 1994 to companies that have them²³. The peer-to-peer platforms had brand and scale at best. When capital sentiment turned, they had nothing holding anyone in place.

What Pari does instead. We operate inside two pre-existing networks with genuinely high exit costs, and we make each one more valuable to its members. This is the subject of the two-layer section below, and it is the core of our defensibility argument.

---

The one that had the right primitive and still died

CircleLending is the most important company in this paper, because it was right about the thing everyone else was wrong about, and it still did not survive.

Founded in Cambridge, Massachusetts in May 2000 and launched in 2001, CircleLending did exactly one thing: it formalised and serviced loans between friends and family²⁴. It explicitly rejected the stranger-lending model that Prosper, LendingClub, Zopa and Kiva would later build²⁴. By 2006 it had raised roughly $10 million from Venrock, Bezos Expeditions and Omidyar Network²⁵. In 2007 Virgin Group took a majority stake and relaunched it as Virgin Money USA²⁶. It had the primitive, the capital, the team, and one of the strongest consumer brands on earth.

It was dissolved in November 2010²⁴. The exit was never formally announced²⁷. Servicing of its existing family loans was handed to a third party. A former executive later founded National Family Mortgage to carry the family-lending model forward in the narrower domain of intra-family real estate loans²⁸.

Why it failed, in our reading.

It sold direct to consumers. This is the decisive error. A family loan is not a product people wake up wanting. It surfaces during a life event (a home purchase, a business, a divorce, an estate conversation), and it surfaces in a conversation with someone the family already trusts. CircleLending had to find families at the exact moment of need through consumer marketing, which is close to impossible and extremely expensive. It had no channel that was already present at that moment.

It positioned as a documentation utility. Papering and servicing a loan is a real service, but it is a one-time, low-price, low-frequency transaction with no recurring relationship. That is not a business that supports venture-scale infrastructure, which is precisely why the successor product narrowed to a documentation service in a single asset class.

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 documented properly. It was alone.

Timing compounded all of the above. Virgin bought in 2007 and the credit crisis arrived immediately; the withdrawal from the US market followed²⁷.

CircleLending proved the primitive and disproved the channel. We took the primitive and replaced the channel.

The distance between CircleLending and Pari is not the product. It is that we do not have to find the family. The advisor is already sitting with them when the question comes up, already knows the balance sheet, already has a fiduciary reason to care whether the loan is structured properly, and has a powerful commercial reason of their own to want the next generation in the room.

---

Obvious white space vs. dense white space

The peer-to-peer generation went after the largest visible population of underserved borrowers. We went after the smallest population of borrowers with the largest amount of capital moving between them. Those are different businesses that look superficially similar on a slide.

The subprime opportunity is genuinely enormous by headcount, and that is exactly what makes it a trap for a venture-backed infrastructure company. 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 in the first place.

Run that comparison honestly and the conclusion is uncomfortable for the consensus view. A platform paying roughly $400 to acquire a borrower who takes one $8,000 loan at a small origination take, with high loss rates and no natural repeat, is running a business that 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 HNW households, for a fixed integration and onboarding cost is running a different kind of business entirely.

Why the demand is real and structural, not speculative

Intra-family lending is not an emerging behaviour we need to create. It is an old behaviour that is currently unserved by software. The National Association of Realtors found that 22% of first-time buyers funded part of their down payment through a gift or loan from a relative or friend in 2025, while the median age of a first-time buyer reached a record 40²⁹. That is not a fringe transaction. It is a mainstream one, executed almost entirely without documentation, interest rate structuring, or tax planning.

At the high end, the transaction is more sophisticated and higher stakes. Below-market family loans have been governed by Internal Revenue Code Section 7872 since 1984, following the Supreme Court's holding in Dickman v. Commissioner that the right to receive interest is itself a valuable property right³⁰. 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 entire structure is exposed on audit.

So the HNW family faces a transaction that is: common, large, legally technical, tax-sensitive, emotionally loaded, and currently handled with a template promissory note and a spreadsheet. That is the definition of a segment where infrastructure creates real value, and it is small enough that nobody built for it, which is exactly why it was still available.

The obvious white space was obvious to everyone, including the people who lost their money in it.

---

Two layers of network effect

Defensibility is not a feature list. It is the answer to a single question: what does it cost someone to leave? For every platform in the mortality table, the answer was nothing. For Pari, the answer is two separate and unusually high exit costs, stacked.

The structural difference is not scale. It is edges. In an open marketplace each node is acquired independently and connects to nothing. In Pari's model, one acquired institution carries an entire client base, and each client carries a family that is itself a durable network.

Layer one: the family network

The family is the most retention-heavy network in existence, and its exit cost is not economic. It is social, and therefore far stronger. A borrower who defaults on a LendingClub note damages a credit score. A borrower who defaults on a loan from their mother damages something they cannot repurchase.

This produces three effects that no stranger marketplace can replicate. Repayment behaviour is governed by obligation rather than enforcement. Participation compounds: a loan between parent and child pulls in the second parent, the siblings who need to understand the equalisation, and often the next generation. And the relationship persists after the loan is repaid, because the family persists.

Pari's product surface is designed around this rather than against it. Documenting the loan, setting an AFR-compliant rate, generating the amortisation schedule, servicing the payments, and reporting the tax position are all services to the family unit, not to an individual borrower. The unit of retention is the household cluster, and it does not dissolve.

Layer two: the advisory firm as an atomic network

The second layer is where the commercial defensibility lives. An advisory firm is close to a textbook atomic network: the smallest set of participants at which the product is fully valuable. Sign one firm and every client of that firm becomes reachable, on the strength of a relationship that firm has already spent years and real money building.

The stickiness here is empirically extreme. Schwab's 2024 RIA benchmarking study puts self-reported annual client retention at 97%¹, and advisory relationships routinely run 20 to 30 years³¹. Even weak firms retain around 92%; the best exceed 97%³². In RIA acquisitions, buyers write client-retention hurdles of 90-95% into deal terms and sellers almost always clear them³³. Clients do not leave advisors. Moving is administratively painful, emotionally awkward, and rarely worth it.

Because clients almost never leave voluntarily, the competition for advised assets is not fought day to day. It is fought at the handful of moments when the relationship is forced open, and the biggest of those is death.

Where the two layers meet: the transfer event

This is the strategic centre of Pari's positioning, and it is worth stating precisely.

Advisor retention of 97% describes the living client. It does not describe what happens when the assets move to the next generation. Cerulli found that more than 70% of heirs are likely to fire or change advisors after inheriting². More recent Cerulli work found that only 27% of future beneficiaries plan to keep their benefactor's advisor, falling to 20% among those who have already inherited. Critically, the reason is not performance. Half already had their own advisor, and 28% simply had no relationship with the benefactor's advisor. Only 10% cited the advisor failing to meet their needs³⁴. In a separate finding, only 13% of affluent investors used the same advisor their parents used, and among the 87% who did not, the overwhelming majority had never even considered it³⁵.

So the largest wealth transfer in history ($84 trillion over two decades, $124 trillion by 2048³⁶) is, from the advisor's seat, not a windfall. It is a scheduled attrition event. Advisors know it: Natixis found in April 2026 that 41% of US advisors regard the wealth transfer as an existential threat to their practice, and 22% say they have already lost significant assets to generational attrition³⁷.

The same research isolates exactly where the leak is. US advisors report retaining assets around 78% of the time when a spouse inherits³⁷, because the spouse was usually in the room. It is the move down a generation that breaks, and it breaks for the same stated reason every time.

And the stated cause of the attrition is the single thing Pari structurally fixes: the heir was never in the room.

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, discussing the family balance sheet, on a matter of genuine consequence to them. Not an annual review they skip. Not an estate document they never read. A transaction where they are a principal.

That is what we sell to the advisor, and it is why the advisor's incentive is aligned with ours rather than merely tolerant of us. We are not asking a firm to adopt a product for a fee. We are offering a firm an instrument that addresses the risk they are most afraid of, using a transaction their clients are already doing informally and badly. When the advisor's retention improves because of Pari, the advisor expands Pari inside the firm. That is the compounding loop, and it runs on the advisor's self-interest rather than on our marketing budget.

Why the two layers compound rather than merely coexist

Each additional family inside a firm makes the firm more dependent on the tooling and more visible to us as a data asset. Each additional firm makes Pari more credible to the next firm, and to banking partners assessing operating history. Each loan enrols additional household members (the co-lending parent, the sibling, the heir) into a relationship that outlives the loan. And every one of those enrolments strengthens the advisor's own network, which is the thing that brought us the family in the first place.

None of the platforms in the mortality table had a loop of this kind available to them, because none of them had a counterparty whose business improved when theirs did.

---

Scorecard

---

What would falsify this

A post-mortem that only indicts other people is a pitch, not an analysis. These are the conditions under which our thesis is wrong, stated as plainly as we can state them.

Banking partner dependency. We do not hold a charter, so movement of funds depends on a sponsor bank or payments partner. That partner will assess operating history, compliance program maturity and capitalisation, and can decline. This is a genuine chicken-and-egg constraint: origination volume builds the history that unlocks the rails, and the rails enable the volume. We are addressing it through compliance investment ahead of scale, a documented BSA/AML program, SOC 2 attestation, capitalisation, and parallel evaluation of alternative rails. It remains our sharpest near-term dependency and we do not present it as solved.

Advisor adoption is slow by design. The same conservatism that produces 97% client retention produces long procurement cycles, compliance review, and reluctance to introduce anything into a client relationship that could go wrong. Signed letters of intent are a real signal but they are not revenue. The thesis fails if firms sign, praise the concept, and never operationalise it. Our conversion sequence from LOI to live implementation is the metric we hold ourselves to, not LOI count.

The transfer-event argument could prove to be a slower sell than the data implies. Advisors broadly acknowledge generational attrition risk. Acting on it is a different matter, and the industry has been discussing this problem for well over a decade without solving it. If firms treat next-generation engagement as a perennial agenda item rather than an operational priority, our strongest commercial argument loses urgency and we fall back to being a useful workflow tool, which is a smaller business.

Family credit risk is lower, not absent. Intra-family loans default. They default less, and they default differently, but a documented default inside a family is a harder event than a charged-off consumer note, with real reputational consequences for whoever facilitated the structure. Our product has to handle modification, forbearance and forgiveness gracefully and with the correct tax treatment, because those are the common outcomes, not edge cases.

Concentration. A distribution model built on a small number of firms is concentrated by construction. Early revenue will depend on a handful of relationships, and losing one is material. That is the trade we accepted in exchange for not paying retail acquisition costs, and it is the correct trade at this stage, but it is a trade, not a free lunch.

An incumbent could build this. Custodians, planning software vendors and large RIA aggregators could add intra-family loan structuring to existing platforms. Our answer is that the compliance surface, servicing operations, and rails integration are meaningfully harder than they look from the outside, and that this is a small line item for an incumbent and the entire company for us. That is an argument about focus and speed. It is not a guarantee.

We would rather write these down now, in public, than discover them in a diligence call. The point of studying a graveyard is not to feel superior to the dead.

---

Notes and sources

---

This paper is for informational purposes only. It is not investment, legal or tax advice, and it is not an offer to sell or a solicitation of an offer to buy any security. Statements about Pari's business model, pipeline and roadmap are forward-looking and subject to change; letters of intent are non-binding and do not represent revenue. Third-party data is attributed in the notes above and has not been independently verified by Pari. Intra-family lending has tax and legal consequences that vary by circumstance; families should consult their own advisors.

The other papers

Analyst briefing How Pari answers the objections in this paper The thesis, the numbers with their sources, and a direct answer to every standard doubt about this category.