The Engine and the Fuel

The Engine and the Fuel

The Engine and the Fuel

How embedded lending turns software into scale, and why the FIDC makes this a distinctly Brazilian thesis

How embedded lending turns software into scale, and why the FIDC makes this a distinctly Brazilian thesis

Reinaldo Coelho and Julia Bertini

Reinaldo Coelho and Julia Bertini

Partners, Triaxis Capital

Partners, Triaxis Capital

May 15, 2026

May 15, 2026

Leitura

Leitura

·

9

9

MIN

MIN

01

The prize of the Finance stage

The prize of the Finance stage

REVENUE LIFECYCLE · ESSAY

Of all the stages of the revenue journey, credit is the one that pays the most, and the one that most reliably destroys those who enter it unprepared. This is the paradox of the Finance stage in the Revenue Lifecycle. Embedding credit on top of software can multiply revenue per customer in ways no other product can match. But credit is not a feature you tack onto a product. It is a balance-sheet operation, with delinquency curves, duration mismatch between the assets you originate and the funding that supports them, and risk that reveals itself with a lag. Treating a credit line as just another SaaS module is the mistake that separates the companies that break from the companies that go on to become some of the most valuable businesses in the digital economy.

This essay is about how to make that crossing, and about why Brazil, thanks to a financial vehicle few markets have, is the best place in the world to make it.

The analogy that organizes all of this is the engine and the fuel. The software provides the engine: the ability to originate credit with an internal intelligence no bank has. The FIDC provides the fuel: the funding that lets that engine run at scale without burning the company's own balance sheet. The thesis of embedded lending is, at its core, the story of how these two couple together.

That credit is the biggest prize in the cycle is no longer speculation, it is data. a16z estimates that embedding financial products into vertical software multiplies revenue per customer by two to five times, while simultaneously reducing customer acquisition cost, because the financial product is being sold to a customer base that is already in-house. But the number that ends the debate comes from Toast, the software platform for restaurants. In a recent year, Toast generated roughly $5 billion in revenue from financial services, versus less than $1 billion from software subscriptions. The tail started wagging the dog. The software, which looked like the business, revealed itself to be the entry point to the real business.

The software, which looked like the business, revealed itself to be the entry point to the real business.

This pattern repeats in nearly every company that makes the crossing. First the software captures the operation, then payments come in, and only then credit, which, when it arrives, often ends up worth more than everything that came before it. It is the empirical confirmation of what the Revenue Lifecycle thesis argues: value concentrates in the deepest stages of the journey, and credit is one of the deepest of them all.

REVENUE LIFECYCLE · ESSAY

Of all the stages of the revenue journey, credit is the one that pays the most, and the one that most reliably destroys those who enter it unprepared. This is the paradox of the Finance stage in the Revenue Lifecycle. Embedding credit on top of software can multiply revenue per customer in ways no other product can match. But credit is not a feature you tack onto a product. It is a balance-sheet operation, with delinquency curves, duration mismatch between the assets you originate and the funding that supports them, and risk that reveals itself with a lag. Treating a credit line as just another SaaS module is the mistake that separates the companies that break from the companies that go on to become some of the most valuable businesses in the digital economy.

This essay is about how to make that crossing, and about why Brazil, thanks to a financial vehicle few markets have, is the best place in the world to make it.

The analogy that organizes all of this is the engine and the fuel. The software provides the engine: the ability to originate credit with an internal intelligence no bank has. The FIDC provides the fuel: the funding that lets that engine run at scale without burning the company's own balance sheet. The thesis of embedded lending is, at its core, the story of how these two couple together.

That credit is the biggest prize in the cycle is no longer speculation, it is data. a16z estimates that embedding financial products into vertical software multiplies revenue per customer by two to five times, while simultaneously reducing customer acquisition cost, because the financial product is being sold to a customer base that is already in-house. But the number that ends the debate comes from Toast, the software platform for restaurants. In a recent year, Toast generated roughly $5 billion in revenue from financial services, versus less than $1 billion from software subscriptions. The tail started wagging the dog. The software, which looked like the business, revealed itself to be the entry point to the real business.

The software, which looked like the business, revealed itself to be the entry point to the real business.

This pattern repeats in nearly every company that makes the crossing. First the software captures the operation, then payments come in, and only then credit, which, when it arrives, often ends up worth more than everything that came before it. It is the empirical confirmation of what the Revenue Lifecycle thesis argues: value concentrates in the deepest stages of the journey, and credit is one of the deepest of them all.

02

Why software underwrites better than the bank

Why software underwrites better than the bank

The advantage a techfin has over a bank in credit origination is not distribution, though distribution helps. It is data. A bank evaluates a small business by what it declares: credit score, financial statements, collateral. A software platform evaluates the same business by what it does: every transaction, every payment received, every seasonal pattern, in real time. Toast, which processes card payments for more than a hundred thousand restaurants, does not need to ask a restaurant what its revenue is. It already knows. And underwriting built on observed revenue, rather than declared revenue, is underwriting of a fundamentally different nature.

The consequence is that the techfin sees good credit where the bank only sees risk. Traditional banks reject the vast majority of small business credit applications, not out of ill will, but because they cannot see the repayment capacity of borrowers who do not fit their standardized models. The software platform sees exactly that, because payment data is repayment capacity, observed rather than declared. This is the same proprietary data moat that shows up when we talk about artificial intelligence: the risk model improves with every transaction processed, and a new competitor has no way to replicate the history. In embedded lending, that moat becomes margin.

Sequence matters, and almost nobody respects it

You cannot originate good credit without having first processed enough payments to know the customer. It is the revenue journey revealing itself in practice: each stage of the Revenue Lifecycle prepares the ground for the next.

Within credit itself, there is a sub-sequence that separates the mature from the hasty. Receivables factoring almost always comes first, because it is short-duration, easy to underwrite, and self-liquidates as payments come in. Term loans and revolving lines of credit come much later, because they require a collections operation, delinquency workout capabilities, and payment flow segregation that few companies have in year one. The temptation to skip this sequence is enormous, because the economics of term lending are so much more attractive. But the operational maturity to absorb the delinquency that comes with it is almost always missing, and that is where companies get hurt. Credit punishes haste.

Credit punishes haste. The economics of term lending seduce; the operation it demands is what separates those who scale from those who break.

The advantage a techfin has over a bank in credit origination is not distribution, though distribution helps. It is data. A bank evaluates a small business by what it declares: credit score, financial statements, collateral. A software platform evaluates the same business by what it does: every transaction, every payment received, every seasonal pattern, in real time. Toast, which processes card payments for more than a hundred thousand restaurants, does not need to ask a restaurant what its revenue is. It already knows. And underwriting built on observed revenue, rather than declared revenue, is underwriting of a fundamentally different nature.

The consequence is that the techfin sees good credit where the bank only sees risk. Traditional banks reject the vast majority of small business credit applications, not out of ill will, but because they cannot see the repayment capacity of borrowers who do not fit their standardized models. The software platform sees exactly that, because payment data is repayment capacity, observed rather than declared. This is the same proprietary data moat that shows up when we talk about artificial intelligence: the risk model improves with every transaction processed, and a new competitor has no way to replicate the history. In embedded lending, that moat becomes margin.

Sequence matters, and almost nobody respects it

You cannot originate good credit without having first processed enough payments to know the customer. It is the revenue journey revealing itself in practice: each stage of the Revenue Lifecycle prepares the ground for the next.

Within credit itself, there is a sub-sequence that separates the mature from the hasty. Receivables factoring almost always comes first, because it is short-duration, easy to underwrite, and self-liquidates as payments come in. Term loans and revolving lines of credit come much later, because they require a collections operation, delinquency workout capabilities, and payment flow segregation that few companies have in year one. The temptation to skip this sequence is enormous, because the economics of term lending are so much more attractive. But the operational maturity to absorb the delinquency that comes with it is almost always missing, and that is where companies get hurt. Credit punishes haste.

Credit punishes haste. The economics of term lending seduce; the operation it demands is what separates those who scale from those who break.

Those who only see the software see a company selling subscriptions. Those who see the entire cycle see a company that will originate credit on its own data and finance it on the best receivables infrastructure in the emerging world. The first reading underestimates what is being built. The second is the thesis.

Those who only see the software see a company selling subscriptions. Those who see the entire cycle see a company that will originate credit on its own data and finance it on the best receivables infrastructure in the emerging world. The first reading underestimates what is being built. The second is the thesis.

03

From product to structure: the Brazilian leap

From product to structure: the Brazilian leap

Up to this point, I have described a thesis that holds worldwide. Now comes the part that is ours. In the United States, embedded lending is a product thesis: the company embeds credit on top of the software and offers it to the customer. But there is a ceiling. Lending consumes capital, and at some point the startup runs into its own balance sheet. To scale origination, it needs funding, and funding, in the U.S., is expensive and typically runs through banks. In Brazil, that ceiling has a way out that few markets offer: the FIDC. And that is why, here, embedded lending can be more than a product thesis. It can be a capital structure thesis.

The FIDC, Fundo de Investimento em Direitos Creditórios, Brazil's regulated securitization vehicle for receivables, is a fund that buys the receivables the techfin originates. Instead of the startup carrying the credits on its own balance sheet, consuming equity capital with every loan, it assigns those receivables to the fund. The FIDC, in turn, is financed by investors who buy its shares. The design is elegant: the techfin originates with its data, the FIDC funds it with investor capital, and the company scales credit revenue without having to scale its own balance sheet.

The regulatory reform of the last few years, through CVM Resolution 175, made this coupling dramatically more efficient. Under the old rule, every new share issuance required an investor assembly, which throttled funds that needed frequent capital raising, exactly the case for receivables factoring vehicles. The new rule allows the fund's regulations to authorize issuances within pre-approved limits, waiving the recurring assembly. And, perhaps most importantly, the reform opened FIDC distribution to retail investors, dramatically expanding the investor base available to finance this origination. What was once an instrument for a few has become market infrastructure.

Abroad, embedded lending is a product thesis. Here, with the FIDC, it can be a capital structure thesis. That difference is Brazil.

The anatomy of the coupling, and its risks

The coupling between techfin and FIDC has a mechanic well known to market operators, and it is worth understanding to see where the risks live. The originator typically holds the fund's subordinated tranche, which is also the piece that absorbs first losses. This is the operator's skin in the game: if the portfolio it originated deteriorates, its capital feels it first. This design aligns incentives powerfully, because the company only scales if it originates well. There is also revolving structure, where the fund uses incoming payments to buy new receivables, creating a continuous origination engine. And the classic risk mitigants are what define fund quality: subordination, overcollateralization, and an investment policy grounded in concentration limits and structured eligibility criteria.

But it would be dishonest to describe the opportunity without the dangers, because they are real and they take down companies. The first is concentration: a portfolio anchored in a small number of obligors or in a single sector carries risk that materializes all at once. The second is duration mismatch: financing long-dated credit with short-dated funding is the classic, fatal error. The third, and the most insidious, is origination quality — the temptation to open the origination spigot to feed the fund is the most common way a good structure becomes a bad operation. There is also the tax weight to consider. Receivables assignment has fiscal implications that need to be priced into margin, even though the FIDC, as a vehicle, is efficient.

None of this is an argument against the structure. It is an argument for executing it with discipline. The FIDC is a precision tool: in the right hands, it multiplies; in the wrong hands, it amplifies the error. The difference lives entirely in the discipline of whoever is originating.

Up to this point, I have described a thesis that holds worldwide. Now comes the part that is ours. In the United States, embedded lending is a product thesis: the company embeds credit on top of the software and offers it to the customer. But there is a ceiling. Lending consumes capital, and at some point the startup runs into its own balance sheet. To scale origination, it needs funding, and funding, in the U.S., is expensive and typically runs through banks. In Brazil, that ceiling has a way out that few markets offer: the FIDC. And that is why, here, embedded lending can be more than a product thesis. It can be a capital structure thesis.

The FIDC, Fundo de Investimento em Direitos Creditórios, Brazil's regulated securitization vehicle for receivables, is a fund that buys the receivables the techfin originates. Instead of the startup carrying the credits on its own balance sheet, consuming equity capital with every loan, it assigns those receivables to the fund. The FIDC, in turn, is financed by investors who buy its shares. The design is elegant: the techfin originates with its data, the FIDC funds it with investor capital, and the company scales credit revenue without having to scale its own balance sheet.

The regulatory reform of the last few years, through CVM Resolution 175, made this coupling dramatically more efficient. Under the old rule, every new share issuance required an investor assembly, which throttled funds that needed frequent capital raising, exactly the case for receivables factoring vehicles. The new rule allows the fund's regulations to authorize issuances within pre-approved limits, waiving the recurring assembly. And, perhaps most importantly, the reform opened FIDC distribution to retail investors, dramatically expanding the investor base available to finance this origination. What was once an instrument for a few has become market infrastructure.

Abroad, embedded lending is a product thesis. Here, with the FIDC, it can be a capital structure thesis. That difference is Brazil.

The anatomy of the coupling, and its risks

The coupling between techfin and FIDC has a mechanic well known to market operators, and it is worth understanding to see where the risks live. The originator typically holds the fund's subordinated tranche, which is also the piece that absorbs first losses. This is the operator's skin in the game: if the portfolio it originated deteriorates, its capital feels it first. This design aligns incentives powerfully, because the company only scales if it originates well. There is also revolving structure, where the fund uses incoming payments to buy new receivables, creating a continuous origination engine. And the classic risk mitigants are what define fund quality: subordination, overcollateralization, and an investment policy grounded in concentration limits and structured eligibility criteria.

But it would be dishonest to describe the opportunity without the dangers, because they are real and they take down companies. The first is concentration: a portfolio anchored in a small number of obligors or in a single sector carries risk that materializes all at once. The second is duration mismatch: financing long-dated credit with short-dated funding is the classic, fatal error. The third, and the most insidious, is origination quality — the temptation to open the origination spigot to feed the fund is the most common way a good structure becomes a bad operation. There is also the tax weight to consider. Receivables assignment has fiscal implications that need to be priced into margin, even though the FIDC, as a vehicle, is efficient.

None of this is an argument against the structure. It is an argument for executing it with discipline. The FIDC is a precision tool: in the right hands, it multiplies; in the wrong hands, it amplifies the error. The difference lives entirely in the discipline of whoever is originating.

04

Where the threads converge

Where the threads converge

Embedded lending through the FIDC is not a financial trick bolted onto a software company. It is the point where the threads of the whole thesis converge. The company enters through software, planted in a vertical, capturing the operation and the data. It uses that data to originate credit with an intelligence the bank does not have — the engine, sharpened by the same AI that amplifies every stage of the cycle. And it uses the FIDC to finance scale without sinking its own balance sheet — the fuel, provided by a financial infrastructure Brazil built and few markets have.

Vertical as the door. Data as the engine. FIDC as the fuel. Credit as the prize. It is the revenue journey traversed all the way to one of its deepest stages, with each piece of the thesis playing its part.

Embedded lending through the FIDC is not a financial trick bolted onto a software company. It is the point where the threads of the whole thesis converge. The company enters through software, planted in a vertical, capturing the operation and the data. It uses that data to originate credit with an intelligence the bank does not have — the engine, sharpened by the same AI that amplifies every stage of the cycle. And it uses the FIDC to finance scale without sinking its own balance sheet — the fuel, provided by a financial infrastructure Brazil built and few markets have.

Vertical as the door. Data as the engine. FIDC as the fuel. Credit as the prize. It is the revenue journey traversed all the way to one of its deepest stages, with each piece of the thesis playing its part.

Authors

Reinaldo Coelho and Julia Bertini

Reinaldo Coelho and Julia Bertini

Partners, Triaxis Capital

Partners, Triaxis Capital

·

Triaxis Capital

Triaxis Capital

Get in touch

São Paulo - SP
Rua Funchal, 411
Cj 64 - Vila Olímpia - CEP 04551-060
Telefone: +55 (11) 3044-4060

Recife - PE
R Sant'Anna, 267
Salas 201 e 202 - Santana - CEP 52060-460
Telefone: +55 (81) 3221-6501

Triaxis Capital LTDA – CNPJ 15.333.310/0001-03 This website is for informational purposes only and does not constitute a public offering of securities. Investment funds are not guaranteed by the administrator, the manager, or any insurance mechanism. Past performance is not indicative of future results. Before investing, read the fund’s prospectus and consult a qualified professional.

Compliance & Regulatory

Get in touch

São Paulo - SP
Rua Funchal, 411
Cj 64 - Vila Olímpia - CEP 04551-060
Telefone: +55 (11) 3044-4060

Recife - PE
R Sant'Anna, 267
Salas 201 e 202 - Santana - CEP 52060-460
Telefone: +55 (81) 3221-6501

Triaxis Capital LTDA – CNPJ 15.333.310/0001-03 This website is for informational purposes only and does not constitute a public offering of securities. Investment funds are not guaranteed by the administrator, the manager, or any insurance mechanism. Past performance is not indicative of future results. Before investing, read the fund’s prospectus and consult a qualified professional.

Compliance & Regulatory

Get in touch

São Paulo - SP
Rua Funchal, 411
Cj 64 - Vila Olímpia - CEP 04551-060
Telefone: +55 (11) 3044-4060

Recife - PE
R Sant'Anna, 267
Salas 201 e 202 - Santana - CEP 52060-460
Telefone: +55 (81) 3221-6501

Triaxis Capital LTDA – CNPJ 15.333.310/0001-03 This website is for informational purposes only and does not constitute a public offering of securities. Investment funds are not guaranteed by the administrator, the manager, or any insurance mechanism. Past performance is not indicative of future results. Before investing, read the fund’s prospectus and consult a qualified professional.

Compliance & Regulatory