A tip off came our way this week regarding estranged video recommendation and discovery software supplier Jinni, with a source close to the matter claiming the company is on the verge of shutting up shop.
Suspicions about turbulent times at Jinni are nothing new as regular readers of Faultline Online Reporter will know, so we were straight on the phone to the company’s CEO Yosi Glick who, rather than engaging in a reasonable discussion about Jinni’s next move in a cut throat market, cited only certain “difficulties” and wanted nothing more than to hang up the call. We are most certainly not in the job of kicking a man while he’s down, simply in accurately reporting significant shifts in the Faultline – and Jinni indisputably has/had its share of significant deployments.
Jinni has been actively seeking a buyer for some time, according to our source, saying an offer was made to purchase the company by video software supplier Ooyala – itself a vendor falling on hard times – around a year ago or maybe more, but the deal collapsed. We have reached out to a few contacts at Ooyala for clarification.
In the meantime, Glick insisted the Israeli vendor still has customers on the books to deal with but was unwilling to elaborate. However, we are a technology publication so the most pressing matter for our readers is where Jinni’s deployment strongholds are and whether these represent potential vacancies. Thankfully, we’ve been tracking Jinni’s movements for years and have painted a pretty decent picture of its major operator customer footprint in the OTT video scene, some of which have never made been public.
Jinni’s chef-d’oeuvre was without question AT&T, where it powered video recommendations across website delivery for a number of years. The rise of in-app video viewing likely squeezed Jinni out of favor eventually at the US operator, although we have been unable to ascertain evidence of a replacement recommendation system sliding in. For us, AT&T has always represented a sizable hole in the customer list of market frontrunner ThinkAnalytics and, over the past four years or so, we have continued to be amazed at Jinni’s ability to retain the AT&T contract in the face of fierce competition. However, we got the impression that Jinni was becoming an ever smaller and insignificant cog within the AT&T machine as in-house developments in the video recommendation field progressed.
Time Warner Cable was another high value deployment for Jinni, a position we saw it struggling to defend alongside established entertainment customers including Universal Pictures, Legendary Studios, AMC Networks and A&E Networks.
Over in Europe and Latin America, Spanish powerhouse Telefonica has been known to use Jinni technology, which provided the recommendation software for the Canal+ system prior to Telefonica’s acquisition of the pay TV unit. Given how Telefonica has kept a tight lid on its recommendation software sources with the exception of Jinni, it was assumed that the system was a firm favorite and lasted past the acquisition of Canal+. However, we promptly pulled this presumption when Italian recommendation rival ContentWise landed a Telefonica deal in April 2018, powering content personalization, discovery and search for Telefónica’s Latin American multiplatform video services – covering live TV, multiscreen VoD and catch-up services.
Indeed, Jinni’s software has historically been dominant in systems built around Ericsson’s Mediaroom, since spun off as MediaKind, which was known to have a gaping hole where recommendation and search should have been.
Belgian incumbent telco Proximus has often been a confusing case, with both Jinni and ThinkAnalytics claiming contracts there. Jinni had a recommendation software deployment at Proximus dating back to 2010, however, in September 2017, ThinkAnalytics published a list of customers which included Proximus and the UK firm has since privately confirmed to us it switched out the Jinni system at Proximus. This might have happened around the 2016 timeframe when ThinkAnalytics added voice activated search. At the time, we gave Jinni the benefit of the doubt, suggesting both sets of software engines may have each been handling different aspects of content recommendation. However, with that scalp confirmed as falling to ThinkAnalytics, there are bound to be skeletons in abundance – with ThinkAnalytics also revealing to us discussions are underway with additional Jinni customers.
Another switch out job was observed at Bouygues Telecom, using Jinni-based recommendations when it first launched Bbox in 2013, but this was replaced with a system from Spideo just a few years later as the French operator developed a taste for delivering content discovery capabilities based on moods and other metrics perceived as more personal. Jinni also holds a US patent for natural language understanding technology.
Jinni, like ThinkAnalytics, gained a reputation for making more compelling and sometimes unexpected recommendations, tuning its engine towards more subtle and advanced techniques than collaborative filtering. Jinni then expanded its product line to a taste-based audience buying platform for advertisers, using its semantic tags to target ads at consumers based on specific tastes and activity. This involved conversion of free user utterances to structured database queries.
A less conventional deployment for Jinni was an integration for its Entertainment Genome in Microsoft’s Xbox games console, inspired by the Music Genome Project – a revolutionary piece of technology at the time in 1999. Elsewhere in North America, Jinni has appeared in set tops from Telus in Canada.
There is no doubt Jinni developed something special in its heyday but clearly some mistakes have proved fatal along the way. “Management would tell me deploying Jinni in a local market (matching all the content to descriptive Jinni ‘genes’) required a large upfront investment and therefore it wasn’t relevant for small OTT players, only big cable companies. I insisted that we look into a way to develop a ‘lighter’ version that could be deployed to the small local/niche OTT companies that would be popping up, and they kind of laughed at me that it can’t be done, and I just don’t understand technical matters. That may be true, but if Jinni could have found a way to be ‘plug-and-play’ for the global OTT market, I think it would have done really well,” elaborated our source.
The fate of Jinni’s intellectual property therefore still hangs in the balance but is of great curiosity to us and our source, who left the company around a year ago but wanted to remain anonymous. “I can attest that discovery is still a huge problem for OTT and Jinni’s solution was incredible. My family is a Netflix-only household and I literally want to cry from the discovery. I find myself wishing out loud for Jinni discovery,” they said.
Our source, clearly still in mourning, isn’t the only casualty, as a quick browse of LinkedIn shows most of the senior Jinni team moved on within the past two years.
Of course, it’s not as if Jinni never had the opportunity to address our growing concerns and talk us through the transition, instead of reveling in clandestineness which only served to arouse suspicions. A couple of years ago, Faultline Online Reporter checked in for catch up with Jinni following an extended period of silence on the announcement front. Instead of giving us an honest update on the business, an over-defensive Glick batted away every question with apathy in almost a carbon copy to this week’s brief exchange of words. The writing was on the wall.