Corelation’s John Landis: The Undisputed King of Core

By John San Filippo

If you’ve worked in credit union technology for any length of time, you’ve probably worked on a John Landis-architected core data processing platform. He’s created three of them, all of which are still in service today. They are, in order of appearance, Fiserv Galaxy, Jack Henry Symitar, and his current venture, Corelation KeyStone.

Finopotamus met with Landis at Corelation’s now largely empty headquarters in the Liberty Station area of San Diego. The facility was designed to accommodate hundreds, but the company went remote during the pandemic and never looked back. Now, except for semiannual company meetings that bring everyone together, only a handful of employees work from the office on a regular basis.

It’s impossible to separate the story of John Landis from the history of credit union core data processing in the United States. To truly understand his impact, one must look at how the evolution of financial technology has mirrored his professional journey, from the era of punched paper tapes to the modern landscape of open APIs and cloud-capable environments.

The Era of Mainframes, Paper Tapes, and BASIC

John Landis

Landis did not set out to revolutionize the credit union movement. Fresh out of college, he was working at Ektelon, a sporting goods manufacturing company owned by his best friend David Held’s father. The business had purchased a Data General computer system. “David’s dad had bought a computer system and we thought, ooh, that would be fun to find out about,” Landis recalled.

The two quickly became self-taught experts on Data General hardware. In the mid-to-late 1970s, computer power was unimaginably constrained by today’s standards. “These were Novas and the early Eclipses,” noted Landis, referring to some of the earliest Data General models. “You had 64K of memory and 10 megabytes of disk space.” At the time, a company called World Computer had introduced the first minicomputer-based in-house system for credit unions, disrupting a marketplace previously dominated by mainframe service bureaus.

A small San Diego data processor called Complete Business Services, Inc. (CBSI) noticed World Computer’s success and decided to shift from its Burroughs mainframe shop into the minicomputer market. A Data General salesman sold CBSI on the necessary hardware and subsequently recruited Landis on behalf of CBSI to build the software.

 “I didn’t know what a credit union was,” admitted Landis. “I had no clue, but I picked it up and wrote them a system that eventually went through two or three acquisitions.”

The technology landscape of the late 1970s was deeply primitive, he added. Some payroll deposit distributions from employers still arrived at credit unions on punched paper tape, requiring a dedicated paper tape reader. Core functionality was straightforward. “Things were pretty simple back then,” Landis explained, “It was basically a teller line, batch ATM, posted dividends, statements.” Automated clearing house (ACH) records and batch ATM transactions were processed via reel-to-reel tapes delivered by couriers.

Landis developed this first system in BASIC because it was the only language available on minicomputers that could support multiple CRT terminals simultaneously. Through successive corporate acquisitions—first by ICCI, then by Citicorp Information Resources—this platform matured and became known as Galaxy (later acquired by Fiserv).

A Meeting at Wendy’s and the Genesis of Symitar

By the early 1980s, corporate consolidation had fundamentally changed the workplace dynamic at Citicorp. Landis brought his friend Held into the company, but frequent corporate restructurings created persistent internal friction. Landis described the management culture as “boss du jour,” adding, “We reported to somebody different about once a month. It was just chaos and not user-friendly to the clients.”

The turning point occurred during a now-legendary lunch invitation at a local Wendy’s restaurant. Manny Prupes, who managed the in-house division, asked Landis to join him in launching a new venture. Landis reflected, “I never would have done it the first time by myself.” He credited his career trajectory to two individuals: the Data General salesman who introduced him to credit unions, and Prupes, who provided the entrepreneurial push.

In the spring of 1984, Landis, Prupes, and Held left Citicorp to found Symitar, establishing equal ownership among the three founders. They hired Theresa Benavidez, who had worked alongside Landis since the late 1970s as a trainer and conversion specialist, as employee number one. Working out of Landis’s home closet and bedroom on a 16-bit computer, he and Held engineered the software from scratch. They converted their first credit union client on January 1, 1985.

In launching Symitar, the founders didn’t focus on bleeding-edge technology as a key differentiator. Credit union cores were still relatively simple, making it difficult to differentiate on technology. Instead, their primary motivation was an unwavering commitment to customer service. They believed that providing superior support and treating clients like family would separate them from large, bureaucratic corporate entities.

Technologically, they migrated from BASIC to PL/I, a compiled mainframe language that delivered the highest performance on Data General hardware at the time. Symitar experienced significant market tailwinds when World Computer struggled with a problematic software rewrite called WorldWorks, which became known in the industry as “World Doesn’t Work” according to Landis, allowing Symitar to steadily capture market share.

Navigating Major Hardware Transitions and Scale Limitations

As the 1980s transitioned into the 1990s, computer hardware underwent a massive structural shift. Proprietary minicomputer architecture was reaching its physical limits. Landis told Finopotamus that core processors of that era were massive circuit boards soldered with numerous integrated circuits, before miniaturization consolidated systems onto a single silicon chip. Data General struggled to exceed clock speeds of 100 MHz, while newer Reduced Instruction Set Computer (RISC) chips quickly bypassed those limitations.

To maintain performance scalability, Symitar decided to look at the RISC-based IBM RS/6000 platform. Landis recalled, “We benchmarked the IBM RS/6000 and went, ‘Holy crap! This thing is so much faster!’”

This shift required moving away from Data General’s proprietary operating system, AOS/VS, to IBM’s variant of UNIX, known as AIX. This transition provided a more standardized environment and drastically expanded computing speeds. Within five to six years, Symitar completely phased out its Data General hardware footprint.

Following the passing of Prupes, the remaining owners decided to sell the company to Jack Henry & Associates in the spring of 2000. At the time of the sale, Symitar had roughly 250 employees and 250 clients. Landis agreed to stay with Jack Henry for five years to oversee critical engineering initiatives.

The most prominent engineering challenge stemmed from an architectural decision made in 1985. Symitar had been built utilizing a 32-bit integer system for financial balances, meaning account values capped out at approximately $21 million. While sufficient in the mid-1980s, this limit restricted scalability for larger institutions. Landis personally engineered the data expansion project, restructuring the fields, transactions, and balances across the system to accommodate asset sizes reaching into the trillions.

Act 3: Corelation KeyStone

By 2005, corporate alignment diverged, and Landis chose to depart Jack Henry. After resting for a few months, he rented a modest 100 square foot office, purchased an IBM RS/6000 server, and began coding the foundation of his third core system, KeyStone. He worked solo for two years before hiring Jeff Dent as employee number one to design the user interface, while Landis focused on backend architecture. Benavidez joined shortly thereafter as employee number two, followed by Landis’ son, Rob, currently the company CEO, as employee number three.

Corelation was founded with both a service-first philosophy and a radically superior technological blueprint. Rather than relying on homegrown databases or proprietary constraints, Landis utilized universal, modern industry building blocks. KeyStone was written in C++ and designed with a pure, browser-based user interface.

While the architecture stack was built to be completely swappable from day one to ensure the platform remained future-proof, it was originally designed for IBM Power Series servers running the AIX operating system. Over time, as Intel and AMD x86-64 hardware and Linux proved to be the platforms of the future, Corelation officially announced support for that environment. Today, all new on-prem conversions run KeyStone natively in this x86/Linux environment with absolutely no difference at the application layer.

Architecturally, the platform was compiled from the ground up to support native multi-threading and uninterrupted 24/7 operations. Landis built extreme structural resilience into the system, ensuring that if power was abruptly severed, the core could simply reboot and resume processing transactions without corruption.

A central feature of KeyStone’s market success is its KeyBridge API. This tool emerged naturally from Landis’s desire to isolate the web-browser user interface from the backend business logic. By building an identical API channel for internal and third-party use, Corelation pioneered an “API-first” framework long before it became an industry-wide mandate.

An Open Playing Field and a Nationwide Workforce

Corelation established an intentionally disruptive business strategy within the financial technology sector: a third-party agnostic approach. Unlike traditional legacy core processors that monetize ancillary applications or charge steep integration fees, Corelation deliberately chose not to charge credit unions or vendors for utilizing the KeyBridge API.

Landis asserted, “We provide the connection. You can connect up anything else you want.”

By refusing to compete directly against specialized third-party vendors for ancillary software packages, Corelation preserves an entirely level playing field. This framework empowers credit unions to seamlessly implement a best-of-breed technology stack without incurring punitive core integration expenses.

This progressive philosophy extends directly into Corelation’s operational environment. Following the disruption of the COVID-19 pandemic, the firm embraced a remote-first organizational model. This strategic pivot lifted geographical hiring constraints, allowing Corelation to source top-tier engineering, conversion, and support talent across the entire United States.

While Landis acknowledged that remote operations lose the spontaneous “water cooler” collaboration of a physical office, the operational gains have been immense. Staff morale remains high due to eliminated daily commutes, and the company reinforces corporate alignment by flying its entire nationwide workforce to San Diego twice a year for comprehensive team-building and strategic meetings.

Reflections on a Lasting Legacy

Today, Corelation services roughly 300 signed credit union clients, managing monumental transitions like the recent migration of Mountain America Credit Union – an institution three times larger than any client the company had previously converted.

As Landis shifts into a part-time capacity to focus more on his personal life, he deliberately avoids immersing himself in new technology cycles. He joked that adding items to a technology bucket list would inevitably sentence him to the grueling schedule he endured during KeyStone’s peak development phase, when he averaged 11-hour days, seven days a week, for 11 straight months.

When reflecting on a multi-decade career that drove the credit union industry from 1 MHz single-core minicomputers to massive multi-core Linux server clusters handling terabytes of memory, Landis remains characteristically modest. He traces his unparalleled career back to a few minor, unpredictable decisions that completely shifted his path.

Beyond the lines of code, compiled databases, and open APIs, Landis occasionally contemplates the human element of his legacy. The corporate cultures he fostered gave rise to lifelong friendships, professional mentorships, and even marriages and families among the employees who helped build his companies.

Throughout it all, Landis has never lost sight of the primary lesson he absorbed from Prupes: Technology is merely a vehicle. True success is dictated entirely by client service. Landis summarized this foundational philosophy cleanly: “The product is only one piece of the puzzle. You’ve got to have a good product, but you can’t have a good product and bad service.”