After years selling money for other people, I decided to turn it into an institution.

Seferu exists to represent companies before the capital market without depending on a single shelf, a single institution, or a single geography.

Last week, Apollo released its second quarter results. The manager now administers more than US$ 1 trillion and originated US$ 317 billion in assets over the last twelve months. Last year it was US$ 309 billion. Roughly 40% of that annual origination came from sixteen specialized platforms Apollo itself built, acquired, or developed around different types of credit.

That deserves a simple translation.

An institution with access to almost absurd quantities of capital spent years building machines to find where to put the money.

Capital, I finally discovered, also has to sell.

I spent almost eighteen years reaching that conclusion by much smaller paths.

In 2008 I still thought of selling as a general skill. The product could change; the capacity to make someone buy survived. In 2009 I heard from an industrialist a sentence I never managed to abandon: nothing makes more money than selling money.

I was nineteen at the time and the sentence looked like a kind of economic provocation. Money would be the perfect merchandise: you hand it over and get it back with a price added.

It took me years to see that the really interesting part was not owning the money.

It was standing between whoever owns it and whoever needs it.

That led me to think about intermediation, liquidity, tenor, risk, collateral, selection, structure, origination, information, and relationships. Little by little it became clear that credit is not simply a price applied to capital. It is an architecture of time and trust. The rate occupies a visible line. The risk is usually hidden in the conditions around it.

In 2019 I started operating that mechanism directly.

Loan broker is the term Americans use naturally. In Brazil the activity appears fragmented across several names, institutional relationships, and specialties. I did not worry much about the name. There were businessmen needing capital and institutions able to provide it. My job was to understand the problem, structure the demand, and bring the two together without turning the client's need into a financial pilgrimage.

I did that low profile over the following years.

It was the best school I could have chosen.

Spreadsheets do not lie by intention, but they omit everything that does not fit the columns. Real operations teach things presentations about credit rarely teach. An economically strong client can be documentally weak. Excellent collateral can be bad inside the wrong structure. A lower rate can cost far more when it requires an inadequate tenor. A famous lender can be terrible for a specific operation. A little-known fund can understand the risk exactly.

I also learned that one of the most dangerous things in finance is the person who always finds a solution.

Sometimes not taking credit is the solution.

If the intermediary is paid only when he gets the money, he has to build enough discipline not to turn his commission into a financial thesis.

That is harder than it sounds.

Incentives do not make people dishonest automatically. They do something subtler: they help honest people find convenient arguments.

That is why I started thinking less as a seller of credit and more as the businessman's representative before the financial system.

That distinction defines Seferu today.

I do not want to build one more product distributor.

I want to build a house.

The businessman arrives with a need.

We start from the problem, not from the shelf.

How much capital does he actually need?

For what?

For how long?

Which cash flow will pay it?

What debt already exists?

Which collateral should be used and which should stay free?

Are there better domestic alternatives?

Do the capital markets make sense?

Private credit?

Some kind of asset-backed finance?

Foreign capital?

Perhaps no debt at all?

Seferu's function is to organize those questions and then represent the company in the market.

That is what being an origination house means to me.

The origin lies in the relationship with the businessman.

The work ends only when the architecture is resolved.

That is different from being a bank.

We do not have to own the balance sheet that will finance each operation.

It is also different from being a platform that hands the client a list of institutions and lets him choose.

The client does not hire us to know names.

He hires us to reduce the number of bad financial decisions he has to make alone.

That is a form of capital advisory.

And it may be the natural evolution of everything I have written since 2008.

For a long time the financial system concentrated product, distribution, and relationship inside the same institution. The manager knew the client because the client kept an account at the bank. The bank owned the products because it owned the balance sheet. The institution manufactured and sold.

That design is fragmenting.

Private credit made that particularly evident.

Managers raise billions and have to originate assets.

Banks keep relationships with companies and can route operations to private capital.

Insurers look for duration.

Family offices participate directly.

Specialized funds carry specific risks.

Capital starts living in places different from where the demand is born.

That increases the value of whoever knows how to connect.

But connecting is a poor word for the work.

A bridge does not analyze what crosses it.

We have to analyze.

There is a company on one side and money on the other. Between them there are information, documents, collateral, structure, currency, compliance, reputation, negotiation, and economic interest.

The operation has to survive all of that.

That is precisely where I started building technology.

Not because I wanted to found a fintech.

Because the work was disorganized.

The client sent documents through different channels. The accountant sent a new version. One institution asked for certain information. Another used a different checklist. The operation's history was scattered across messages, emails, files, and memory.

While there are few deals, the professional compensates with attention.

When origination grows, attention becomes the bottleneck.

The best financier on the team starts spending time looking for a PDF.

That is an almost offensive use of human capital.

The obvious solution seemed to be turning the system into software and selling it to other originators.

For a while I considered exactly that.

Today I think it may have been a small reading of the asset we had.

If an engine can increase the productivity of origination, there are two ways to monetize it.

I can charge a subscription to whoever wants to use the engine.

Or I can use the engine to execute more and better operations and participate economically in the result.

I chose the second.

Seferu's software stays closed.

Not because software is secondary.

Precisely because it became too important.

It is our proprietary infrastructure.

The client does not have to learn to use it.

He may not even have to know everything that exists behind the desk.

He sees people.

Behind the people there is a machine.

Company, documents, history, deals, lenders, proposals, collateral, pending items, commissions, capital intelligence, and memory come to exist in the same system.

Every operation leaves something behind.

The next time that company needs capital, it does not start from zero.

We already know its story.

That repetition changes the business profoundly.

The first deal is credit.

The second starts being a financial relationship.

After a few years, Seferu may know that business family's balance sheet better than any of the individual institutions it works with.

That is the point where origination starts approaching a family office.

Not because I want to manage the family's investments.

Because capital does not end at the operation.

A family has a company, real estate, liabilities, financial assets, collateral, and plans.

I cannot seriously advise on the R$ 20 million debt without asking what the R$ 200 million around it are doing.

The view of the whole improves the decision about the part.

That is one of the reasons I like the analogy with Turim.

Turim does not have to manufacture every fund, bank, or instrument it uses. Its economic position comes from sitting next to the family and representing the wealth before the financial suppliers.

I want to build something similar on the corporate capital side.

A house that sits next to the businessman.

Banks compete.

Funds compete.

Structures compete.

The main relationship stays with us.

That does not mean antagonism toward banks.

Quite the opposite.

Banks are extraordinarily important suppliers of capital and will frequently continue to be the best solution available.

Independence does not consist of running from the bank.

It consists of being able to choose it when it is better and refuse it when it is not.

The entire financial system can become a supplier.

That is the game.

And it gets especially interesting when we add international capital.

I spent the last two years thinking more seriously about that.

There is a growing quantity of private capital looking for assets beyond traditional markets. Not all of that money wants Brazil. Not all wants the middle market. Not all accepts reais. Not all accepts the legal and currency risks of emerging markets.

I have no interest in pretending any Brazilian company can now phone Apollo.

That is access marketing.

Real access starts with understanding when an operation fits.

For some companies, the best capital will remain Brazilian.

For others, mainly larger operations, assets with revenue in hard currency, acquisitions, asset-backed structures, or specific situations, widening the universe can radically change the terms available.

The company should not have to know the answer in advance.

That is our function.

If foreign capital makes sense, Seferu has to be able to make the company legible to it.

Translating the presentation is not enough.

We have to translate the risk.

Corporate structure.

Beneficiaries.

Financial statements.

Indebtedness.

Cash flow.

Collateral.

Currency.

Hedge.

Governance.

Data room.

Compliance.

The foreign lender is under no obligation to believe the family "is very strong in the region."

It has to understand why.

That requires a quality of preparation that can improve domestic access as well.

Companies frequently pay a spread on disorganization without knowing it.

The creditor cannot price what it does not understand, so it has to protect itself.

The question nobody answered shows up later in some form.

In price.

In collateral.

In limit.

Or in refusal.

One of Seferu's functions is to reduce that ignorance.

Not to eliminate risk.

To make the real risk more visible.

That is good for both sides.

A lender should want good, legible companies.

A good company should want its risk understood correctly.

When the two meet, there is a deal.

Apollo helps visualize the scale of that logic.

In 2025, it originated US$ 309 billion. Roughly 40% came from sixteen platforms specialized in different types of assets, from mid-market lending to aviation, equipment finance, real estate, trade finance, and other verticals.

They are not sixteen pieces of software.

They are machines for producing assets.

That difference matters.

Apollo does not sell its origination infrastructure so third parties can make money using it.

It uses the infrastructure to feed its own capacity to put capital to work.

That reinforced a conviction I had already been forming.

Seferu should not sell the engine.

It should become the machine.

Naturally, our scale bears no comparison to Apollo and anyone who tries to suggest otherwise deserves some suspicion.

The principle, however, is transferable.

Origination is an asset.

Technology can increase its productivity.

Data produced by the operation itself can improve the next operation.

Relationships with capital can compound over time.

The market a house can access can become a moat.

That is what I intend to build.

A boutique small on the outside and much larger on the inside.

Few senior people in the client relationship.

Technology absorbing the repetitive work.

Artificial intelligence used where it really reduces toil, not as homepage decoration.

A proprietary system learning the map of capital.

And a culture disciplined enough to keep saying no.

That last point matters more than any feature.

Seferu cannot become a factory whose need for volume forces it to find debt for any company.

That would merely be a correspondent in better clothes.

I want to build reputation on both sides.

The businessman has to know we will not sell his need to the first institution willing to pay.

The lender has to know we will not turn its attention into a dumping ground for deals.

That limits short-term growth.

It probably increases long-term value.

Reputation works that way.

It is slow to accumulate and fast to destroy.

A bad operation can pay very well.

It can also consume a capital relationship worth ten good operations.

The fee shows up in the income statement.

The reputational damage does not.

That is why financial houses need a longer horizon than their salespeople.

Seferu is born with that problem solved in a particularly simple way: it is mine.

I have no need to satisfy a venture capital investor expecting user growth because I do not want to build a business of users.

I want to build an institution.

The number that interests me is not how many people logged in.

It is how much capital we organized correctly, for how many companies, with what quality of execution, and how many of them came back.

Volume matters.

Recurrence matters more.

A client who comes back means the first operation created a relationship instead of merely a commission.

That is the business.

The first door may be credit.

Then we start knowing the company better.

Perhaps an acquisition appears.

A refinancing.

An asset sale.

An international structure.

Private capital.

Perhaps a situation in which neither debt nor equity alone solves it.

The house grows with the client's complexity.

That strikes me as far more interesting than trying to sell every product from day one.

We do not have to be everything.

We have to be the first call when capital becomes a decision.

That position has enormous value.

That is what I learned watching bankers for years. A good banker changes institutions and sometimes part of the relationship goes with him. Not because he stole the client. Because the account belonged to the bank; the trust did not.

I want Seferu to institutionalize that trust without turning it into dependence on a single person.

Software helps precisely there.

History cannot live only in my head.

If the house depends on me to remember every detail, I did not build an institution.

I built a sophisticated job.

The memory has to belong to Seferu.

Documents.

Decisions.

Previous operations.

Collateral.

Lenders.

Results.

The rationale.

At the same time, I do not want an organization whose answer is the product of fields filled in on a screen.

Context still matters.

A family company is not simply a tax ID, EBITDA, and LTV.

There are people.

Objectives.

History.

Concentration.

Risks numbers do not explain.

The system keeps facts.

The financier has to keep interpreting.

That is the division I want to preserve.

Technology for memory and repeatable intelligence.

People for judgment and responsibility.

Perhaps in ten years the frontier will be different.

I do not have to know today.

There is enough work in front of us.

Brazil has hundreds of thousands of relevant companies served mainly by fragmented financial relationships. The bank sees its own exposure. The accountant sees numbers. The lawyer sees structures. The businessman keeps an enormous part of the context in his head.

There is room for an institution that organizes capital around him.

We do not have to serve everyone.

In fact, we probably should not.

I want companies where the complexity justifies advice.

Business families.

Private companies with a history.

Operations relevant enough for structure to matter.

People for whom a few points of rate matter less than avoiding the wrong collateral, the wrong tenor, or the wrong counterparty.

Seferu has to be able to be a boutique.

That means accepting that scarcity is also part of the strategy.

I do not want thousands of small credit requests thrown into an automatic funnel.

I want to know companies.

I want to understand why they need capital.

I want to build a history.

In the short term, that probably generates far more revenue per client than software.

In the long term, it can build something far harder to replace.

Software can be copied.

Relationships, proprietary execution data, and reputation with capital are not copied at the same pace.

That is the asymmetry.

Every operation pays us today and, if we do it right, also increases our capacity to operate tomorrow.

That is business compounding.

A company served produces a relationship.

An operation produces data.

A lender produces history.

History improves matching.

Matching improves execution.

Execution increases reputation.

Reputation brings better companies.

Better companies attract better capital.

That is the cycle I want to feed.

Not ARR.

Not active users.

Not the number of features.

Seferu may one day have some of those indicators internally. They should not command the business.

In recent years, many people learned to call any financial company with software a fintech.

The word does not interest me.

Seferu has technology.

It does not have to sell technology.

It has access to capital.

It does not have to be a bank.

It provides a service.

It does not have to work like a handmade consultancy.

Perhaps the classification stays uncomfortable.

Good.

Interesting institutions frequently appear before a comfortable name exists for them.

For now, the definition that suffices for me is simple.

Seferu is an independent house of origination and capital solutions for companies and business families.

We organize the demand.

We structure it.

We prepare it.

We take it to market.

We negotiate.

We coordinate through to closing.

The capital can come from banks, funds, securitization companies, private credit, family offices, or, when appropriate, foreign sources.

We do not promise the money.

We promise to represent the problem correctly before whoever has the money.

That difference is enough to build a reputation.

Perhaps one day a very large business as well.

In 2009, I found the idea of selling money fascinating.

Today, after years doing exactly that, the sentence lost its shine and gained a better definition.

Money is not the product.

The decision is.

The company pays us so that the need for capital is turned into a decision that preserves as much of its future freedom as possible.

The rest is instrument.

That is the institution I decided to build.

Its name is Seferu.

Leo Bentier