Guidewire Labs, co-founded by The DEMO conference executive producer Chris Shipley, plays matchmaker between startups and enterprises with its Fusion platform. Behind Fusion stand features such as natural language processing of companies’ submitted business and product essay descriptions and ontological trees that pair discrete industry-standard descriptions with discovered terms.

“We like to think of ourselves as a matchmaking service like eHarmony, but only for business opportunities,” says Stephen Pieraldi, Guidewire Labs COO. “Companies want to be matched but it’s very hard to do that, unlike with people where you have basic filters. But no one has taken a profile object called company and identified it in the way we have.”

Why should company matchmaking migrate to the digital semantic web? Manual doesn’t scale. Imagine a conference like LeWeb for its startup competition, for which it uses Guidewire Fusion: “In the past, first and foremost they would have to enroll companies and handle the customer service of getting their forms,” says Pieraldi. Then that information would have to be manually administered and entered into a back-end database, then sorted for distribution to reviewers to make their painstaking way through it to choose winners.

“There’s no semantic search, no matching, no relevance or context. So in the old days LeWeb would take 60 to 90 days to process its application load. In our system it takes two weeks,” he says.  “It’s time management and streamlining the process.”

Working in Fusion

With Fusion, startups can build a universal profile of critical business information and share their information with potential partners, investors or other sources within a secure, transparent environment. The universal profile, the company says, allows startups to enroll easily in corporate innovation programs, competitions, and sponsored events, while maintaining control of their business information. They can individually control which of their attributes enterprises may or may not have access to for searches, so they always know their contextual relationships to partners, Pieraldi notes, and they can opt out of relationships as well. Enterprises can follow those they are interested in collectively and manage interactions through one console.

The proprietary G-score assessment methodology in Fusion, a National Science Foundation backed business fingerprinting framework, provides a means for companies to do self-assessments across a number of categories that helps with making matches with enterprises looking for particular expertise or other capabilities. “For example, some large acquiring companies would like to make a match with a company that is technology-driven vs. market- and sales-driven,” says Pieraldi.

The service boasts 5,000 companies fully sharing profiles and some 10,000 sharing limited profiles in its database, he says. “So when you make a request you are matching against thousands of companies,” Pieraldi says.

As a simplified example of its semantic smarts at work, think of an enterprise that wants to build a relationship with companies that create Teflon rolling pins. A standard system search could uncover lots of companies that make rolling pins, and even Teflon ones specifically. But with Guidewire Labs, for a company that may have put non-stick rolling pins into its description but not Teflon, Fusion will be able to figure out that equates to Teflon and so include the company in the list of returned results.

“It’s really profile matching,” he says, “The interest is within the product description and background…if they said non-stick rolling pin, because we implant it as a leaf in the tree of the semantic ontology, we can say we know you asked for this, but five companies have an occurrence of Teflon even if they may not say Teflon.” From there enterprises can narrow things further by exploring in what other ways the potential partner meets its requirements – whether it has enough revenue to support an order request, what level of expertise its leadership has, and so on.

Guidewire Labs soon will be adding new capabilities to its service. Among them will be the ability for enterprises in search of innovation partners to build their own requests rather than have the vendor do it for them; a get opportunities feature that provides a list of companies proposing events for a company to join; and search results that return not only explicit companies that match enterprise criteria, but others that seem to be worth talking to based on what Guidewire knows about it – even though it can’t specifically tell the searcher why. (For example, the startup might be in stealth mode and subject to certain constraints about the information it can disclose.)

Further down the road, Pieraldi says, the service likely will incorporate peer-to-peer searching, as well.