Prior Knowledge made it to the finals last year at TechCrunch Disrupt SF. They did not “wow” the judges, but their predictive database technology for developers certainly turned attention to what they do.
Three months later, Salesforce.com acquired Prior Knowledge for an undisclosed price. And then Prior Knowledge shut itself down — permanently. They killed their flagship API, erased all the user data, shut down the website and erased all traces of their work on GitHub.
The decision to shut down the service and go into super secret mode came as Salesforce began to put the pieces together for its marketing cloud and the reinforcement of its developer platform. But they lacked a crucial piece — the ability to do sophisticated predictive analytics in a way that made it easier for customers to see connections and make decisions.
Enter Prior Knowledge, which had developed what it called the Veritable API, a service that Co-Founder Eric Jonas had said gives developers super powers to build powerful applications with predictive analytics baked in. Salesforce.com does provide analytics to drill down on sales and social media data. It does not, though, have the capability to build a graph in the shape and form of a service like LinkedIn, which I wrote about last month:
Since its start 10 years ago, LinkedIn has become the place for people to network. In recent years, though, it has started pooling the data, becoming one of the early adopters of open-source data technologies, such as Hadoop and Lucene/Solr for search. It has one of the most-recognized teams of data scientists who have learned to shape the data to create what CEO Jeff Weiner calls a global economic graph. It’s through the understanding of its users’ interactions that LinekdIn is establishing a platform that could put it in a position to emerge as an enterprise services provider and a player in the CRM market.
A year later, Jonas is keeping quiet about the work he and his team are now doing at Salesforce. Jonas, who is now chief predictive scientist at the company, declined an interview for this story, only hinting at some news to come for the work his team is doing. Here’s what he said in an email about his new role: “I work on discovering applications for our technology, work on research prototypes, and scientifically guide the team that works on predictive analytics and machine learning.”
Jonas was working on his doctorate in neurobiology at MIT when he took a leave of absence to work on Prior Knowledge. He and his team of statisticians, engineers and mathematicians received $1.4 million in funding from The Founders Fund to develop a service that basically modeled the knowledge of the team into the technology itself. It is reminiscent of Wise.io, the machine learning as a service that harnessed the knowledge of a team of astrophysicists from the University of California at Berkeley. Co-Founder Joshua Bloom said at their launch that he and his colleagues at Berkeley would often manage data that was well beyond what conventional tools could offer. They had to invent the technology themselves. Now all that knowledge is getting packaged to offer as a service.
A Few Clues
There are clues scattered about that tell a story about the direction Salesforce is taking. Beau Cronin was a co-founder at Prior Knowledge. He is now a senior manager of predictive products at Salesforce and contributor to O’Reilly Radar. In a post last April he wrote about the need for robust analytics and essentially the importance of Prior Knowledge:
I find the database analogy useful here: developers with only a foggy notion of database implementation routinely benefit from the expertise of the programmers who do understand these systems — i.e., the “professionals.” How? Well, decades of experience — and lots of trial and error — have yielded good abstractions in this area. As a result, we can meaningfully talk about the database “layer” in our overall “stack.” Of course, these abstractions are leaky, like all others, and there are plenty of sharp edges remaining (and, some might argue, more being created every day with the explosion of NoSQL solutions).
Making the point how the database has become abstracted sums up what the Prior Knowledge team set out to do. And that’s simplifying the complexity of adding deep analytics into apps. That’s a daunting task for most developers these days. Salesforce has an aggressive mobile strategy. Much of their work has historically focused on making it a bit easier to develop apps. The latest update of its mobile platform points to this effort. Making analytics easier to implement would dovetail with Salesforce strategy.
Furthermore, Salesforce has made some substantial investments in its marketing cloud. Predictive analytics for mobile marketing is a gold mine that Salesforce is surely seeking to tap.
The secrecy I encountered points to the value that the Prior Knowledge technology has to Salesforce, which has invested heavily in building platforms for app development but has left predictive analytics pretty much out of the equation. The Salesforce Data.com platform has a degree of analytics but it is primarily marketed as a sales lead database. The lacking predictive analytics capability is a missing piece, pointing to a weakness in the company’s overall product strategy that the Prior Knowledge technology should help shore up. But even more so it is the people from the team who can direct the scientific path for Salesforce to follow so they might find that data gold mine.
Oh, and for your reference, here is Jonas presentation at last year’s event. Not exactly the most compelling of presentations but certainly enough to get attention for the power the technology offers.