‘Reverse engineering’

Like many of the largest companies, Merrill Lynch can’t always buy the IT tools it wants. Sometimes they just don’t exist. That’s why four years ago the financial services firm started building its own software to expose mainframe resources to other applications via Web services.

The firm went the homegrown route after unsuccessfully searching for a packaged product that met its needs, says Andy Brown, chief technology architect. Nothing surfaced at the time or over the next few years. “We never found any compelling reason to stop doing what we were doing.”

Until now. The software product Merrill Lynch built, called X4ML, is going commercial through its new owner, SOA Software , which plans to announce the deal this week.

“We felt like we’d invested quite a bit of money in the technology already, and we also felt like the opportunity in the open market was such that [the product developers] needed to be given a chance to take their idea and test it in the open market,” Brown says about the sale of X4ML to SOA Software.

The commercialization of technology that got its start in an enterprise IT setting, such as X4ML, is not unprecedented. There’s a constant flow of intellectual property from government agencies and universities to the commercial world. Companies in other industries have done the same thing.

Shell funded Kalido, a spinoff formed in 2000 to tackle data management challenges. Boeing spawned MessageGate, a messaging security and compliance vendor, in 2003. And newly launched BlueNote Networks was inspired in part by Fidelity Investments’ desire for VoIP-based collaboration software.

Most recently, start-up TrueBaseline came out of stealth mode and announced a product based on technology developed by Westinghouse Electric. Stringent regulatory requirements for reporting changes to IT system settings led Westinghouse to develop its own tools when commercially available configuration management products wouldn’t suffice, says Aruna Endabetla, CTO at TrueBaseline and a former project manager at Westinghouse.

Westinghouse used the product internally for several years before marketing it to other companies a couple of years ago. But trying to expand the product to meet wider demand, as well as execute software sales, proved incompatible with its core business. Westinghouse decided to sell the IT assets to TrueBaseline’s founders.

The independence is a good thing, Endabetla says. Before the formation of TrueBaseline, developers in Westinghouse had to balance the nuclear electric power company’s needs with those of the broader market – which sometimes slowed down evolution of the technology. “Now we can look at the real market needs and develop the product for the entire market,” she says.

Seeds of invention

Companies today seem to be thinking harder about how they can leverage non-strategic technology assets, says Bob Hower, a general partner at venture capital firm Advanced Technology Ventures. “We’ve had conversations with several companies lately that are trying to figure out how to get more mileage out of the intellectual property they’ve developed.”

But the transition from in-house technology to commercial product isn’t always successful. Sometimes poor personnel choices can derail a spinoff. “Not every large corporation IT person is an entrepreneur,” Hower says.

In addition, not every nifty IT invention is suited for broader markets. A product that originates inside an enterprise IT department “often has too much local culture,” says Bill Gassman, research director at Gartner. There can be patent issues, scalability shortcomings and code problems. “It may have become spaghetti code by the time it’s ready to go public because feature after feature has been added to the product, and the discipline of commercial software development just wasn’t there.”

In general, there’s often valuable technology inside companies that has the potential to be commercialized, but most companies don’t have the resources to accomplish it, adds Vern Brownell, CTO at blade server vendor Egenera .

Prior to founding Egenera in 2000, Brownell was CTO at Goldman Sachs. During his 11-year tenure at the financial services firm, Brownell considered the commercial potential of some of its technology.

“A number of times we had what we thought was pretty good technology,” he says. But readying an in-house product for use outside the institution required a level of investment that didn’t seem justified. “There were other higher-priority strategic or competitive things to work on from a development standpoint that made more sense.”

What came out of Brownell’s experience at Goldman Sachs was the idea for Egenera. Brownell had helped shift the financial firm’s infrastructure from predominantly mainframes and midrange servers to a heterogeneous, distributed computing environment and knew firsthand the management complexity that resulted.

Brownell didn’t see anything on the horizon to alleviate the problem – but he thought there needed to be. That thought led to Egenera.

“Customers have unique perspectives that vendors don’t always see,” Brownell says. “When you’re in the middle of it, when you wake up in the middle of the night with cold sweats because of what you’ve created, you have a different sense of urgency about it.”

Balancing act

Having a former IT user at the helm – such as Brownell in Egenera’s early days – can help a start-up get off the ground, Hower says. “It lends credibility to a team,” he says.

It also helps when a vendor’s technology has been shown to work in a demanding enterprise IT environment. In the case of the X4ML technology, “if that has been made to work in a financial environment where high reliability, security and performance are all big issues, it probably has a better chance of being successful” than other technology spinoffs, Gassman says.

SOA Software says the Merrill Lynch technology will help extend its product portfolio to the mainframe. Enabling mainframe resources to participate in a services-oriented architecture (SOA ) has been a challenge, says Eric Pulier, executive chairman of SOA Software. “Getting these CICS mainframes to become part of the new infrastructure has been problematic – problematic for us to deliver to our customers, and problematic for customers to find a solution.”

The deal also gives SOA Software critical expertise: A team of four Merrill Lynch programmers and engineers who built X4ML are going with the product to SOA Software.

Such a staff shift is usually unavoidable in the process of commercializing in-house technology through a spinoff or technology acquisition. It’s likely that there will need to be some reworking of the product for the wider market, and “that can be hard to do without folks who know where all the skeletons are,” Hower says.

On the other hand, for the company that developed the intellectual property, a spinoff can mean losing key personnel, Gassman says.

For Merrill Lynch, making the decision to sell the X4ML technology required trade-offs, Brown says. “You want to do the right thing for the people who created the technology, the right thing for Merrill Lynch shareholders and the right thing for the future of the technology itself.”

Not everyone at Merrill Lynch initially agreed that the X4ML technology should be sold. While Brown had no hesitations, he says it was not as easy to convince others in the financial services firm. “Executive management is often uncomfortable with divestment because they’re wondering what they’re giving away,” Brown says.

Convincing everyone – and then choosing the right buyer from among “quite a few companies who were interested in buying the software” – took about six months, Brown says. In the end, management agreed and everyone walked away happy, he says.

“The folks that developed the code are very happy that they’re now able to go look at the real addressable market instead of the addressable market inside Merrill Lynch.”