I spent Tuesday and Wednesday largely offline at the 2005 In-Q-Tel CEO Summit. As those in the VC world already know, In-Q-Tel is the strategic venture fund of the US intelligence community. Lacking the scale economies that allow the DOD to have a captive research arm, DARPA, it was the CIA that originally funded In-Q-Tel as a way to gain early access to venture funded technologies critical to the intelligence mission. The program has accomplished just that, as well as scoring some exits on its early investments. It was reauthorized in 2004, when other relevant agencies, such as NGA, joined the CIA in support. (Disclosure: Pacifica Fund has one co-investment with In-Q-Tel, Bay Microsystems.)
The conference is an interesting mix of portfolio company executives, mid-level and senior intelligence agency personnel, and VCs and other hoi polloi. For a VC, it's a chance to look at companies which have already had their technology and capabilities vetted by one of toughest and most sophisticated customers in the world, and which are already receiving cash flow as a result. Your mission (should you choose to accept it) is to figure out which of the ventures also have scalable, fast growth opportunities on the civilian side.
An unsurprising interest of the agencies and focus of the In-Q-Tel portfolio is natural language processing. As I have a nearly two decade long personal interest in text and unstructured data search, organization, and discovery, the conference is an opportunity to catch up on the state of the art, as filtered by those with extreme mission-critical needs. In-Q-Tel and the intelligence community are also beginning to push these best-of-breed vendors into tighter integrations, such as a collaboration between Arabic OCR vendor NovoDynamics and Arabic/English machine translation experts Language Weaver. Since I agree with the remarks of panel moderator Ron Weissman of Apax Partners, to the effect that the unstructured data landscape has too many vendors without critical mass, here's hoping this kind of collaboration will naturally lead to some rollup activity in the field. While the intelligence community has the expertise to match technology to requirements down to the project level, that sophistication is not matched on the civilian side. Law firm IT departments (to pick one example) aren't able to differentiate between many of the offerings, leading to a fragmented, sales driven market and limited exit opportunities. Can we have a little M&A music, please?
Many two day investor conferences feature a 'best of show' vote and panel as a wrap-up. That would be a bit tactless on In-Q-Tel's part, since all those presenting are already part of the family, but there's nothing stopping me from dishing out my own awards. So, I hereby present the Due Diligence awards for:
- Coolest Technology, to Cambrios, for its genetically based approach to creating nanostructures that will be useful in small scale circuits, materials, and sensors. In an example given in the presentation, the genome of a bacteriophage was randomly modified and the resulting organisms screened for selective binding of the protein coat to GaAs. This one's still in the science project stage, but it doesn't take too much imagination to guess the relevance to things like sensing of biochemical threats. It's also an example of bootstrapping nanotech from the inherent scale economies of natural, evolved systems, something that deserves another post.
- Best Civilian Application, to Call Miner. The VC community has seen plenty of speech recognition plays, mostly centered around off-the-shelf engines used in call center applications. replacing IVR systems and live operators. Corporate and consumer acceptance has been spotty, shall we say. But here's a new twist: Look at speech as a discovery and analytics target, where even low accuracy word spotting can result in useful insights on shifting content and timing patterns of customer interactions, diagnosing response issues within a call center as well as noticing shifts in inquiries triggered by service or product problems. No points given for guessing the intelligence applications of this one, but here's hoping the 'customers' involved receive the world's best high explosive products, express COD.
- I Want It, to Destineer for First to Fight, a game based on a Marine Corps training simulation. Fire team level tactics built from the POV of those who have been there. This is out on the Mac as well as PC and console platforms. Looks like it will melt the CPU on my poor old TiBook. Agency applications:
Play testing Coffee breaksSimulation. Yeah, that's it!
- Best Keynote, to Mark Leslie for a talk featuring his concept of the Sales Learning Curve, an experience curve for the customer facing portion of a venture, analogous to the well known learning curve in manufacturing. Highly recommended for venture investors and early stage managers, particularly those entering the go-to-market phase of the company. Update: You can find a version of the presentation here (PDF). Hat tip: Dave Chase.
Thanks to the staff of In-Q-Tel, who put on a crisply run event that has the best tchotckes you'll find at an investor conference. Where else do you get a Spy Pen with your registration materials? Last year's boodle included a CIA cookbook (did you know Julia Child was once an agent for OSS, the CIA's predecessor?) , and this year's featured Spy-Fi, a print version of the collection of fictional spy props currently housed at CIA HQ. Both well received on the home front. Intelligence is a deadly serious business, but these guys have some tongue-in-cheek fun along the way.