Three years ago a friend introduced me to Palantir Technologies, which heavily recruited software engineers and system administrators for their bread and butter finance product. They have since expanded into the nonprofit and health industries. Today at the 10th Girl Geek Dinner, Palantir Technologies showcased a product that links missing children, registered sex offenders, reported runaways, witness sightings, demographics, and geospatial databases for the National Center for Missing & Exploited Children (NCMEC).
Chitra Ragavan, advisor to Palantir Technologies, brought to the attention of CEO Alex Carp an opportunity to expand into non-profit space. Based on Chitra’s investigative report on missing children, “Lost and Found” in the U.S. News and World Report, she advocated for starting with NCMEC. In a DC conference, Linda M. Krieg, currently the NCMEC Assistant Executive Vice President of Operations soon to be Chief of Operations by the end of the year, made a hard sell to Palantir. Without hesitation, Palantir very much wanted to contribute to NCMEC. The sell was mutual.
NCMEC measures itself by the number of cases it solves. With mounting missing children cases, they needed all databases to talk to each other and narrow down leads. Analysts spend much of their time manipulating data by merging and querying databases through MySQL, Access, Excel, SAS, and other database manipulation programs. And for several government entities, a matter of piecing together faxes. Palantir leaps forward with an intuitive UI that can take in voluminous data and crunch it within milliseconds. When AMBER Alerts go out, witness reports and sightings are taken as text reports. A smart highlight of keywords are used to query relevant databases. For example, partial license plate descriptions can be used to filter automobiles registered to known delinquent sex offender parolees. The list can be further narrowed down by the height, hair color, and other appearances or notable blemishes that characterize the suspect. Analysts spend less time constructing a query and more time analyzing the output provided by Palantir.
Later that evening I had the pleasure to speak with Melody Hildebrandt, who leads Palantir’s 3-person health group. Without much success responding to government request for proposals, Palantir directly approaches government clients that champion technology, such as Todd Park, HHS CTO. Todd’s campaign for Open Government has lead to the Health Data Initiative Forum, which Palantir presented with their CDC data visualization UI.
They also teamed up with the Center for Medicare & Medicaid Services to link Medicare expenditures with hospital quality and provider location. This is helpful in analyzing geographical variation in provider practice patterns driven by the risk mix of the patient population and their level of access to necessary and appropriate health care services.
Palantir is also going after electronic health records to help analysts interpret interesting historical patient and medical relationships that usually get brushed aside due to time constraints. Palantir puts in the muscle work in performing real-time data manipulation and output. Data manipulations is no longer an analyst’s arch nemesis for time. They can now mark off their wish list of issues to explore.
On a side note for business pricing strategy, Palantir bills by CPU core, as a function of data volume and query complexity. It is not clear how much development costs (fixed FTE developer salaries) and capital overhead affect their bottom line. Palantir has the incentive to have a well-defined product that prevents scope creep. The more tailored the product is up front, the easier it is to prove product effectiveness. Palantir has incredible confidence in the quality of their products, and it shows based on receptive clients and partnerships. This sells itself to the client to further trigger additional features driven by crunching more data. This financial model keeps giving and feeding itself. Brilliant business pricing to compete with large consulting firms that rely on hourly or fixed budget fees.