VentureBeat

Ten companies presented during today’s morning portion of Techcrunch40 conference in San Francisco. If we were able to have stock in these companies, here’s how we’d rank them: CastTV, Cubic Telecom, Yap, Cognitive Code, Viewdle, Powerset, Trutap, Faroo, Ceedo and Loudtalks. Summaries of each company follow.

CastTV, the video search engine — We’ve written about the company before, and it still hasn’t launched publicly. The development today is that it showed live demonstrations for the first time. It showed how a search for someone like “Britney Spears” yielded better results than Yahoo, Google or YouTube. For example, unlike YouTube, which showed primarily dated videos of Britney, CastTV showed more recent videos of Britney, and a mix of things like her in movies, or dancing, or in the news. Even Marissa Mayer, Google’s top product executive, one of the panelists chosen to ask questions of the companies, conceded that its interface, timeliness and clustering were strong. Not that she had any choice. CastTV showed a search for “Colts Titans” on Google video, and it showed a video of last year. CastTV has a game from yesterday. CastTV is different because it searches all sorts of surrounding code on pages to tell it what a video is about. It also indexes videos that may not have unique URLs. This may not become the biggest company of the bunch, but is likely to be bought by one of the bigger players — and so a good bet.

Cubic Telecom, a mobile company for international calling — Essentially, cheap overseas calls from anywhere, or at least cheaper than most calling plans. Their service, MaxRoam, is clearly well planned-out. For a more in-depth analysis, check out VentureBeat’s separate post today on the company. Competing mobile VoIP services are cheap or even free, but are unreliable. Its offering you a chance to take your regular phone oversees, without worrying about hassles.

Yap, speech recognition software for mobile phones — Yap’s application is aimed primarily at people who use text messaging. When the user speaks into their phone, Yap instantly translates what they say into a text message (although, in their demo, the translation was noticeably laggy). Aside from instant messaging, Yap also connects to various services, including Twitter, Digg, Wikipedia and commercial sites like Amazon and Ebay, sparing the user the need to type queries or messages. The text messaging service is monetized through suggestive text ads. For instance, if the user mentions a movie in theaters, Yap will suggest a theater to see it in. Although Yap comes with an all-star cast of developers who have worked on projects for AT&T and Apple, the question is whether the big mobile carriers will be interested enough to include Yap software on their phones, or whether the market will be restricted to people who download it themselves. There are other recognition technologies, so the risk is whether it will be adopted quickly enough.

Cognitive Code, offering conversational artificial intelligence – This company is newly launched. During its brief demo, the company’s executives asked questions of “Sylvia,” an artificial intelligence software that answered in a female voice to questions such as “Please close the Word file.” This is very early, but it was impressive that Sylvia seemed to understand some lengthy conversation questions (she’s doesn’t try to merely understand key words, but also the context provided by various words together). Sylvia worked half the time, but failed the other half. She’d do things like open documents instead of closing them as asked. It’s hard to see how much depth the company has, without someone not related to the company trying it out. The company wants to embed the service in other applications. The company says it wants to target 2009 (during CES conference) for hitting the market, and is aiming to embed the technology in toys and games.

Viewdle, a facial-recognition site for video search engines — This company recognizes images and faces in videos around the Web. It advances on what companies like Polar Rose are doing for images. The demo was notable. If you search for Britney Spears, the engine churns out a list of clips as a results. Select one of the clips, and it will open directly at the point in the video where Britney appears, even if only momentarily. More interestingly, the demo showed that if you searched for well-known models, it would pull up videos where all three appeared. You can then zoom in on one of the models, and search for videos only of her. It also provides names for other people it recognizes in images. The big question, however, is how the engine will recognize anyone who is not a star or famous. It counts on folks like you and me sending video links of ourselves, so that it can put us in the database.

Powerset, the semantic search engine we’ve written about before — The company is developing a way to understand the meaning search queries. So if you type in “What do politicians say about Iraq,” it provides results that don’t necessarily relate to those exact words, such “politicians.” For example, one result is an article with the words “President Clinton explains Iraq strategy.” The only new development today, however, is that Powerset announced three demo sites, including “Quotes,” “Business,” and “Powermouse.” The latter gives you insight into Powerset’s semantic database, so that you can see what things it associates with people or things. President Clinton for example, is a politician, lives in the White House, sets policy, etc. The question is whether people will use Powerset, when Google is so good.

Trutap, for mobile social networking — We’ve also written a separate post about this company here. It offers social networking over mobile phones. Features include messaging to groups, text messaging, instant messaging, and blogging in a reasonably straightforward phone application. Launches Friday. Although Trutap can rank xx place in this lineup, its future as a company is as hazy as any of the upcoming social networks. It’s looks functional, but the demo didn’t blow me away. Their focus seems distinctly British. During their presentation, the CEO sent a message to a “Friday Night Crew” planning a visit to a pub. Although it’s not an unreasonable scenario for American users, small social differences can mean popularity in one market, but failure when translated to another.

Ceedo, a lightweight visualization platform for mobiles — The term “visualization platform” doesn’t exactly inspire excitement, and neither did the company’s product. The platform transfers self-contained desktop environments from computer to computer; users can, for example, plug their USB dongle into a rented computer at a cafe and have their home software available. Ceedo has offered such software, for enterprise or personal use, for some time. Their product launch at TechCrunch is a similar platform for mobile phones. The software allows users to run their mobile phone platform on their computer, sending messages or downloading music. Or they can run their home desktop environment on a different computer by plugging in their phone. Although Ceedo’s platform appears to run seamlessly, it just doesn’t strike us as a must-have — and since their job is to convince mobile carriers to include their application on cell phones, that may be a problem. It has plenty of competitors, too.

Faroo, offering a peer-to-peer search engine — The service lets people download an application, so that they can contribute their searches and computer power to a shared search engine. Faroo follows what people search for, what result pages they look at, how long they they look at them, and whether they bookmark them. That way, they use people’s actions to rank a page higher in results. The company’s founder says that one million people using it can index 10 billion pages, meaning the entire Web. The big question, is who on earth would use this? How does it seed the engine with results so that they are relevant from the get-go, without forcing the early adopters to wade through painful early steps and making pages relevant? It is based in Erkrath, Germany. Long shot.

Loudtalks, an “internet walkie talkie” — Perhaps it was just bad luck, but Loudtalks’ presentation fell flat, between technical difficulties and the thick accents of its (apparently) Russian founders. The company’s computer software offers a way to instant message your friends with voice, speaking to them just as you would if you were sitting next to them. You can do this whether you’re on a computer or mobile phone. Loudtalks could prove to be a good idea, but numerous pitfalls spring to mind, among them having a hundred buddies simultaneously blabbing at you. It’s notable that while Loudtalks focuses on simply conveying a user’s voice, other services like Yap work to translate speech into text. It’s often most convenient to give the receiver a text message, even if the sender would prefer to speak. Finally, the voice market is crowded. Why would you download a separate program for this, when there are so many alternatives you’re already likely to be using.

(This was co-written with Matt Marshall.)