Adam Carmel wanted a better answer from his software. He was running a mortgage company, trying to price loans, and finding that the systems meant to help him were getting in the way. An unsatisfactory product can produce a complaint, a workaround, or a resigned renewal. Carmel eventually produced another company. There is a certain justice in the customer becoming the competitor, although the customer must then discover how much work goes into answering everyone else’s complaints.
That passage, from Ethos Lending to Polly, explains more about Carmel than a funding announcement can. He had operated the business whose tools he wanted to replace. He knew the mortgage quote from behind the screen: the margins, the eligibility rules, the requests for changes. His ambition grew out of those details. Seven years after founding Polly in 2019, he is still pursuing the consequences of that early dissatisfaction.
An education with unfortunate timing
Carmel studied at the University of Southern California between 2003 and 2007, earning degrees in business administration and political science. His professional life then supplied a rather pointed sequel. He entered the mortgage industry in 2007, just before the credit crisis. As first jobs go, it offered limited opportunities for complacency. A newcomer learning how the business worked was also watching the business discover what could go wrong.
His early employer was Provident Funding Associates, where his career included a vice president role. In his own account of those beginnings, he described being hired at a real estate investment trust before moving into mortgage work, including servicing. That matters because servicing follows the loan beyond its sale to the borrower. It is an introduction to the continuing operation, rather than only the satisfaction of getting a deal across the line.
He says he started his own mortgage company in 2013. Ethos Lending’s company profile gives the same founding year. Looking back, Carmel has been willing to describe himself as too young when he took that step. It is a pleasingly economical admission. Starting a lender can look quite different after one has actually had to run it.
Ethos combined residential lending with an explicit interest in the production process. It reported $5.5 billion in cumulative mortgage originations and set itself a goal of reducing the cost of manufacturing a loan by 90 percent against incumbents. The first number describes loans made; the second describes an ambition. Together they show the setting in which Carmel began asking more demanding questions of his tools.
Pricing rules, eligibility and margin decisions sit behind the number a borrower sees. Carmel encountered them as a lender before selling software to lenders.
The customer crosses the counter
The pricing engine was the particular irritation. Carmel’s account is practical: he decided to build one internally. Then he went to other mortgage executives and asked whether they had encountered similar problems. He says they had. Only after those conversations did he take the wider opportunity to his board.
The sequence is worth lingering over. A tool built for one’s own operation has an immediate test: someone has to use it to do the work. Asking peers introduces another test: whether the problem travels beyond the building. Neither guarantees a viable company. Both are useful protection against mistaking a founder’s private annoyance for a market.
The board supported the idea. Carmel later explained that the mortgage company’s principals sold the business and that some proceeds helped capitalize Polly. The software venture began in 2019. An operational problem had acquired a separate commercial life, along with the much larger obligation of supporting other people’s operations.
That distinction gives the story its bite. A former customer can remember precisely what it felt like to wait for a vendor. Once he becomes the vendor, customers can ask him to live up to his own argument. Every request for flexibility arrives with an implicit reminder: this is why you said the company needed to exist.
A little less waiting, please
In 2022, Carmel’s public argument centered on agility. Different lenders have different strategies. A community bank, a credit union and a large independent mortgage company do not necessarily want to configure their work in the same way. Software has to accommodate those differences if the customer is to remain in charge of the business.
He also argued that end users belong in the development process. “Collaboration is key,” he said. It is an agreeable phrase with a demanding implication. Listening means that the roadmap must survive contact with a customer who has a deadline, an unusual requirement and very little interest in the elegance of the vendor’s original plan.
Polly’s integration work supplied a concrete illustration of how much lies beneath the promise of convenience. Carmel described connections with other systems taking anywhere from roughly two weeks to a couple of months, depending on the counterpart. Testing and running test data through the connection were part of that process. The finished screen may look simple; getting two businesses’ software to agree is its own occupation.
His public writing has returned to the same territory. Articles published under his name discuss automation, application programming interfaces, mobile interfaces and the distinctions between cloud technologies. The common concern is how lenders actually work. The writing makes an extended case for paying attention to the machinery beneath the experience.
Money for the machinery
Polly’s January 2022 announcement put a financial figure alongside that case: a $37 million Series B led by Menlo Ventures. The company said its customer count had nearly tripled during the preceding year and included several of the country’s top 100 lenders. These were company-reported measures, but they gave the young business a more substantial setting than an interesting prototype.
In September 2024, Polly announced another $25 million in growth equity, led by 8VC with participation from existing investors. It reported that lock volume had grown by more than 150 percent that calendar year and that its customer base had more than doubled. The money was intended to support research, products and expansion, including customer implementation and service.
Those last functions deserve a place beside the engineering. A lender changing software needs the replacement to become part of a working operation. A sales agreement is a beginning. Implementation is where the promise starts encountering the customer’s actual day.
Carmel’s year-end reflection on 2024 acknowledged growing pains in 2023 and early 2024. He also emphasized the team and culture. Recognition followed his work: he was named a HousingWire Vanguard in 2024 and appeared among Inman’s Best of Finance winners in 2025. Awards add punctuation to a career. The work still has to resume the following morning.
Announced equity rounds. Financing amounts, not sales or personal wealth.
“We’re going to fight for every inch to make that happen.”
Adam Carmel, on reducing mortgage costs
The answer arrives before month-end
On October 6, 2026, Polly introduced PollyOS, combining pricing and eligibility, hedging and loan trading on a common data foundation. The practical question was loan-level profitability: what happened to the expected margin as a loan moved through the operation? The system was designed to keep that answer connected to the same loan record.
The distinction between an expected result and an eventual one can sound like accounting housekeeping. For a person running a lender, it changes the usefulness of the information. A number available while decisions are being made has a different job from a number assembled after the month closes. Carmel’s earlier frustration with visibility now had a broader product expression.
The new Hedge system and the reintroduced Loan Trading Exchange extended the company’s reach beyond the original pricing task. Polly said changes including cancellations and funding would be reflected in the pipeline in real time. These are product claims about how the platform works, with a clear purpose: give operating teams and executives a consistent view sooner.
There is a personal continuity here. Ethos put Carmel in the position of wanting better tools. PollyOS puts him in the position of explaining what a connected set of tools should make possible. The scale of the product has changed. The demand for an answer a lender can act on remains recognizable.
Learning to use the brake
In an October 2026 conversation, Carmel was asked what the previous seven years had taught him. He joked that the subject deserved a longer session and “a couple bottles of wine.” Then he described a lesson about pace: knowing when to move quickly and when to slow down. The remark gives a little welcome texture to a career otherwise narrated through very serious software nouns.
He credited advisors including Rob Kessel and Gary Malis, alongside customer feedback, with helping shape the work. Kessel founded Compass Analytics, an earlier participant in mortgage capital markets technology. Bringing that experience into Polly’s next chapter makes the project a conversation across generations of tools, with people who understand how the earlier decisions came to be made.
Carmel’s argument for moving more carefully was readiness. The team wanted to listen, learn and get the system into a condition it could stand behind. Speed remained an ambition, but it now had to share the room with judgment. A founder who had objected to the limitations of existing software was taking responsibility for the timing of his own.
His stated goal remains reducing costs and helping lenders improve profitability. Whether those benefits are retained, competed away or passed to borrowers depends on decisions beyond a software launch. Carmel has chosen to work on the part he can build: the information and controls inside the operation. The story began with a dissatisfied customer at a pricing screen. It continues with that customer on the other side of the counter, expected to provide a useful answer.