There’s no such thing as a free lunch, especially when it comes to software engineering. Low-code tech providers, though, are looking to challenge that axiom.

With the low-code application market set to reach nearly $50 billion by 2026, its place in the upper echelons of financial services is starting to take shape. Low-code platforms, and their sibling no-code platforms, proposition something almost too good to be true: that anyone can quickly build their own applications with minimal to no coding expertise. The premise is contradictory to the ongoing contest between Big Tech and High Finance to hire the best and brightest engineers.

To be sure, low-code and no-code software contains code. It’s just that the user side need not bother with it, as apps are built visually, or often “Lego-style” by custom dragging and dropping pre-configured components and widgets onto a blank slate.

Every new programming language is more or less built on top of older ones. In programming’s humble beginnings, there was machine code (binary 1’s and 0’s) and assembly code (a low-level, plain-text version of machine code). These are most easily digested by a processor. An app written in a high-level code—say, Java—will get compiled down to machine code in order to run on a processor. No-code and low-code are extensions of this, which creates an additional level of abstraction that makes software building easier, faster, and more accessible to those with non-programming backgrounds.

Software is a time-consuming, expensive, error-prone kind of skill that people—especially in the enterprise—struggle to do efficiently, so you end up with big banks spending billions of dollars on it. And so, there’s always been this kind of thinking—is there a shortcut? Is there a way we can just simplify this?

Jon Butler

Mark Beeston, founder and partner at venture capital firm Illuminate Financial—a leading investor in low-code platform provider Genesis—says that the low-code movement has been slow to take hold in the capital markets because there are standards for resiliency, latency, and security—all of which are filtered through the oversight of an up-close regulatory eyeball. It’s an industry that, by its very design, is complex and has no silver bullets.

That said, he takes the view that low-code and no-code have their part to play in the evolution of financial technology, marking a distinction between evolution and revolution.

“It’s a good enabler for an industry that is still heavily reliant on huge-scale deployed software with huge-scale on-prem data storage, that has spoken a good game about cloud migration, big data, AI and all these other things, but in reality is still in the very, very early stages of its adoption of those,” Beeston says. “The Lego-block approach to microservices infrastructure, and the low-code product generation that can sit on top of that framework, is a big enabler of successfully going through that journey.”

Taking Sides

Proponents of low-code and no-code “engineering” assert that it circumvents legacy vendor technology, solves the buy-vs-build dilemma, and fosters an army of citizen developers, each of whom double as business experts and technologists. They also posit that, as coronavirus-induced remote working further takes root, leaving on-prem hardware solutions and tech departments often miles away from users, the practice will quickly go from accessory to necessity.

Across the aisle, skeptics contend that good software takes no shortcuts, and that there’s a trade-off between ease and error, especially at the enterprise level, making a project more time-consuming and expensive to maintain in the long run.

“Since software engineering was invented, everyone has tried to find shortcuts to try to make it a faster process—make it a more deterministic process,” says Jon Butler, chief executive of Velox Financial Technology. “And then every attempt when you do it at scale, and you do it over a long period of time, the corners that you cut to get to something quickly, usually you end up paying for in the end.”

Despite mounting buzz, those same skeptics argue the best use-cases for low-code and no-code remain simple, manual tasks, and workflow-like approval chains—useful in the same ways that robotic process automation (RPA) is useful.

However, the tech is relatively young—though that, too, is up for debate—and sources say that as adoption grows, so will its propensity toward more complex systems, such as trading technology and data delivery. The central question, then, is how much disruption, if any, do the low-code and no-code styles of development pose to finance—and indeed, to software development as a field?

Lay of the Land

Before low-code and no-code became what they are today, they had a different name. Growing up in the 1980s, they were once known as CASE technologies—a tidy acronym for computer-aided software engineering—or expert systems, which are logic-based computer systems that solve problems through “if-then” rules. They garnered hype and praise in their own right, before peaking in the 1990s, with businesses citing frustration over rigid adherence to design conventions and restrictiveness in developing complex, custom applications.

Now seemingly back from the dead—perhaps not a surprise for 2020—several vendors in the space are starting to make some noise. Five-year-old Genesis launched its low-code application platform specifically designed for capital markets in October of last year, following a $3 million Series-A funding round led by Illuminate Financial and Tribeca Early Stage Partners. This past May, it joined the Fintech Open Source Foundation (Finos).

Rival company, three-year-old Unqork, which raised over $50 million in February to expand its platform, boasts an enterprise-grade no-code platform used in a bouquet of industries, including financial services, healthcare, education, real estate, and government. It recently teamed up with consulting giant Deloitte on a number of initiatives, including small business lending and broader financial business challenges, and counts Goldman Sachs among its investors.

I would say the word ‘developer’ should be eliminated, because the word itself actually has no meaning. Developer means someone who writes code; it doesn’t mean they actually solve the problem.

Gary Hoberman

And of course, there are the established tech providers looking to pounce on a growing market. In June, Amazon Web Services (AWS) jumped aboard, rolling out its own tool—though not specifically aimed at finance—called Amazon Honeycode, which allows users to build web and mobile apps without writing any code. Two weeks later, New York-based alternative data analytics provider Apteo released Predictive Insights, its own no-code predictive analytics platform, a move which Apteo’s CEO, Shanif Dhanani, hopes will expand the company’s client base beyond financial services.

Even investment banking behemoth Morgan Stanley has thrown its hat in the ring with its open-source project Morphir, which allows the bank’s users and developers to automate software development by mapping computer science techniques to real-world business problems—for example, the definition of an investment-grade bond. The tool generates its own code representing that core logical concept, and that code can be implemented within another platform, regardless of the codebase in which that platform is written.

With an assortment of startups, tech giants, and financial services firms entering the fray, it’s easy to see why so many believe that this is a market that is ripe for growth in the coming years.

Same Space, Different Philosophies 

“I would say the word ‘developer’ should be eliminated, because the word itself actually has no meaning,” says Gary Hoberman, founder and CEO of Unqork. “Developer means someone who writes code; it doesn’t mean they actually solve the problem.”

He says this is evidenced by the percentage of enterprise tech projects that fail. A 2015 publication by Boston Consulting Group found that 93% of large IT projects with budgets greater than $10 million failed to deliver value to their organizations. A more recent paper published by market research group Forrester Consulting found that out of 315 US, UK, and German firms—financial firms made up 16% of that figure, the largest demographic—35% of projects failed to meet their original business intent, 38% failed to meet initial timelines, and 34% failed to finish within budget.

Unqork’s platform doesn’t allow users to override any of its own functionality unless it’s programmed with the ability to be turned on or off—a stipulation that, in a certain light, seems to run contrary to no-code’s draw of customization and vendor freedom, though it does ensure the service stays as close to its code-free promise as possible.

Unqork’s first clients were in the paper-intensive insurance sector, building no-code user interfaces—wizards, which typically require thousands of lines of code be built—for companies like Liberty Mutual Group, Principal Financial Group, and the Hanover Insurance Group. Then, the company entered banking and capital markets, with Goldman Sachs as its first client.

Before making its first $22 million investment into the startup, the bank rolled out its first tool powered by Unqork last year. The solution collected applications and data from employees at the bank wanting to apply for an in-house incubator called GS Accelerate, which in 2018, received 1,000 applications.

“We became a system of record there, not just a workflow system,” Hoberman says. “We not only could be a front-end to your trading platform—your EMS, your OMS, your clearing, your reconciliation—but in addition, we will be the core system; we will be the EMS and OMS. We will be the actual high-speed transaction to the Street.”

Genesis co-founder and CEO Stephen Murphy has similar thoughts of Wall Street domination. You can use Genesis for simple business process management (BPM) use-cases—and indeed, this has been a large focus for customers, Murphy says—but there is a budding interest from firms who want to use low-code to automate more complex financial processes.

In July, European bank ING partnered with Genesis to launch its Credit Insurance Application (CIA) offering, a deal-flow and portfolio application intended to provide connectivity and digitization on the credit and political risk insurance market. CIA is the bank’s first successful production deployment using any low-code development platform, says Matt Rhys-Evans, director of loan market innovation at ING.

A simple solution for CIA could have been creating a basic deal-reporting database, but as the bank’s “squad,” which consisted of business members and tech staffers in its London-based Innovation Lab, documented the whole internal and external user process for arranging credit insurance, they pitched the idea of building a customizable, cloud-hosted solution to Genesis instead. The team broke down the workflow into nine steps, delivered Genesis a data model, and the vendor designed a minimum viable product, which handles the majority of the bank’s transaction management as well as monitoring and controls. Collaborating on the entire build was about 2/3 to 3/4 technologists, and 1/4 to 1/3 business and design members, across Genesis and ING.

“CIA replaces a large spreadsheet that we were using across the business to record deal activity, and this was prone to errors, incomplete data, and reconciliation issues—all the usual issues with overdependence on a spreadsheet solution. Not only that, people would normally process certain tasks by way of their own file extracts and deal templates before integrating data in the portfolio spreadsheet. Users can now connect through CIA, sharing transaction data securely, complete the majority of tasks inside the application and work in real-time with a complete picture of the portfolio, live checks and controls, and activity visualization,” Rhys-Evans says. “It’s a massive improvement.”

On top of using Genesis’ low-code, pre-built modules, ING did need to perform some extra development work on top to accommodate for complexities and a new, bespoke use-case. But Rhys-Evans says he looks forward to being able to use only the drag-and-drop method—“making it even faster,” he says.

The low-code movement has been enabled and empowered by the rises of microservices and application programming interfaces (APIs). While a monolithic enterprise application might be made up of a database, a user interface, and a server-side application, and available through a web app, a microservices architecture connects the capabilities of a business to users via open APIs.

Because these are built in standardized formats like REST, they provide a means for software systems to “talk” to each other and facilitate integrated apps. APIs can be bundled into architectural frameworks like API gateways, which act as a single point of entry for multiple APIs.

Microservices also serve as the foundation for Murphy’s Genesis. Beginning as a microservices infrastructure stack for capital markets, the company began positioning itself as a low-code application platform only about two years ago (it launched its full-fledged low-code application platform, LCAP, in October of 2019, prior to which it had released low-code and no-code web-based tools).

But there’s a striking difference between the likes of Unqork and Genesis, one perhaps rooted in practicality versus purity.

With Genesis, you can write your own code. Genesis will continue to take care of the innermost pipes—like how low-latency messaging will work between microservices, for instance, or complex real-time event monitoring—while users can enhance and customize their front-ends with additional code. The platform is also equipped with code libraries containing Python, Java, Angular, React, and other languages and libraries for programmers to utilize.

In the next month or so, Genesis will announce the completion of a complete new trading platform built on its service, combining business-process-management workflows, new trading technologies, and a lot of legacy integration using both in-house and third-party apps, Murphy says. It will be the first of its kind for the platform, but it signals a certain maturation.

‘Everything, Squared’

A chief technology officer at a US investment bank says people have been talking about these solutions for years, over which time he has formed the opinion that they’re a good choice for workgroup applications, but no one would seriously consider using them for any enterprise application.

“For instance, DealCloud was originally built on Sharepoint—which could be thought of as a low-code platform—but eventually, they had to migrate to a real platform. Also WordPress is another low- or no-code platform for websites, and it is terrifically popular, but at some point if an application has legs it will outgrow these platforms, and then they will need to be migrated to a more performant [sic] platform,” the exec says.

In terms of financial data, it’s very hard data because the signal-to-noise ratio is very low. … If you’re moving closer to the trading side of the business … I’m not aware of any kind of mainstream system that can easily do that.

Valery Manokhin

Jon Butler concurs. Four years ago, Butler left Goldman Sachs, where he was a managing director for 15 years. He went on to spend a little more than a year as the head of capital markets technology at Deloitte, before co-founding Velox in 2017, where he currently resides. Though he left the bank before it formed its relationship with Unqork, he says Goldman had been considering low-code and no-code solutions for years prior with little success.

As it relates to trading functions, he is still skeptical of the technology. A software engineer by trade, he understands why others might call him a “dinosaur” in his thinking, and he understands why those same people would want to move away from traditional programming—but he doesn’t think low-code and no-code are completely viable alternatives, at least not on a big scale, just as each new programming language written doesn’t replace those of yore—just look at the fact that Cobol is still quite prevalent inside of numerous banking systems.

“Even 50 years ago, software was everything. And obviously now, software is everything, squared,” Butler says. “Software is a time-consuming, expensive, error-prone kind of skill that people—especially in the enterprise—struggle to do efficiently, so you end up with big banks spending billions of dollars on it. And so, there’s always been this kind of thinking—is there a shortcut? Is there a way we can just simplify this?”

When banks are spending millions, or billions, on a tool or platform, they won’t settle for something that “just about works,” Butler says. At Goldman, end-users of the bank’s low-code and no-code early experiments often wrote and re-wrote around the edges of the tools—what Butler calls the movement’s “back door” problem.

Back doors, or custom configurations—a capability Genesis offers, but Unqork does not—is where trouble often started. On one hand, it allows the creator to tailor the tool to their specific wants. But on the other, it opened it up to security and control issues, and the organization took on risk.

“You create a bit of a monster [this way],” Butler says. “In the old world, the technology department builds the technology. They have controls around it, they have processes around figuring out where the code is, version control, and security, and all this stuff. With low-code [and] no-code, anybody can write anything, so what you end up with is thousands of these apps that become critical to processing, but aren’t really fully controlled.”

Valery Manokhin, a machine-learning researcher in probabilistic forecasting at Royal Holloway, University of London, sees both sides of the debate, being a coder himself and having experience with the different ways to use code in business and academia. He says there’s a decent amount of diversity in low-code and no-code tools, as well as in what they can accomplish, and cautions against generalizing them and what they can do.

Softwares like Tableau (acquired by Salesforce in August 2019) and Microsoft’s Power BI are low-code development platforms built for conducting analyses or product back testing using visualization. And they’re very good at doing that, Manokhin says.

“Say a business wants to have a dashboard of how, for example, a trading model is performing, or a factory wants to a see how the machines are performing—its’ a great tool, right? But this is only really a summary. It doesn’t do anything intelligent. This is why I always say this is only the first step to take a pulse of where the business is going,” he says.

In his work as a lead data scientist for a Fortune 500 company, Manokhin has currently been focusing on open-source development projects using Python and R libraries to create forecasting models centered on global operations and supply chains across the company’s footprint. He uses Power BI at the end to visualize the forecasts with “the click of a mouse.” But he has also recently started using another low-code platform, Alteryx, which he says has yielded some pleasant surprises when compared to similar offerings for performing predictive analytics—if only because the system lets data scientists like Manokhin code on top of it.

However, low-code may not mix as well with investment data, which changes quickly and is plagued by rapid alpha decay, says Manokhin, who has technical specializations in quantitative finance and systematic trading.

“In terms of financial data, it’s very hard data because the signal-to-noise ratio is very low. That’s why it’s very difficult to predict,” he says. He views back-office management and customer service as promising homes for today’s low-code tools, but “if you’re moving closer to the trading side of the business, if you want to be more intelligent … maybe even design algorithms that could do things better in terms of trading, I’m not aware of any kind of mainstream system that can easily do that.”

At Odds

Interestingly, the low-code and no-code movement is adjacent to, and in some respects, directly opposes another trend in capital markets, one that arguably has an even stronger footing—the desktop app interoperability movement.

If low-code and no-code aim to give users the freedom to build their own user interfaces, the method of the three interop kingpins—OpenFin, Glue42, and Cosaic, formerly known as ChartIQ—is to keep the complex interfaces of users’ third-party apps, and make those UIs talk to one another.

Dan Schleifer, co-founder and CEO of Cosaic, says low-code and no-code have a promising pitch, but even as users build their own interfaces, they can’t totally get away from those of their third parties, and the ongoing maintenance cost of a low-code system grows over time.

“Usually folks only build out a tiny fraction of the user interface for a given app—say, a few bits of Salesforce—and then the user still needs to go back to the main vendor UI for anything but the most basic of functions, such as advanced search, long-tail features, etcetera,” Schleifer says. “Not to mention, as the vendors innovate and deliver new functionality, you then have to go in and replicate it in your own bespoke low code UI.”

He says the use-case around order and execution management systems (OEMS)—a complicated system that Unqork says it can serve as—is interesting, however. Firms often end up with a mish-mash of OEMSs and they want to pull them together into a single interface (this problem also is giving way to the rise of server-side interoperability). But, Schleifer says, it’s unclear how well it works to accomplish this in reality, and whether it simply further entrenches what the OEMS users have and don’t like.

It’s possible then, that the question isn’t whether low-code and no-code are good or bad, or whether they work or not. It might become: how compatible are they with finance—is this a hammer-looking-for-a-nail scenario?

It’s easy, then, to recall a very different craze that received a lot of hype but very little production—“the 2016 blockchain gold rush, where it was like, ‘Hey! We’ve got a technology!’ Great. Now show me a problem,” as Illuminate’s Beeston characterizes it, though Beeston does believe that, in this case, there is a problem to be solved using low-code applications.

The bank CTO says something similar, which is that any software development platform will provide code users don’t have to write in the form of libraries to use, and low- and no-code can be thought of as just being further along this spectrum—it’s, essentially, all part of engineering evolution.

With so many firms already in the space, and with so many opinions about the need for low- and no-code platforms, perhaps the real question is, where exactly is the industry on the Gartner hype cycle: Is the movement heading toward the peak of inflated expectations; is it about to fall into the trough of disillusionment; or is it finally making its way up the slope of enlightenment.

There’s no agreement here. Perhaps it’s a conversation to be had over lunch.