Business intelligence company Pi—formerly PI and
before that PI Benchmark—has launched a "travel cost-analysis platform"
to consolidate transactions from numerous data sources and sort them by trip.
CEO Keesup Choe told BTN the tool,
named Self-Learning Automatic Trip Engine, uses machine learning technology
similar to Google Photos' use of facial recognition for tagging photographs.
SLATE interrogates data feeds from travel management
companies, card providers and expense reports and attempts to figure out which
costs belong in the same trip. Pi claimed the tool helps clients understand
their true total trip costs, leading to improved supplier negotiations and a
better appreciation of their returns on investment in travel.
The machine learning element fills in incomplete data
by, for example, connecting an ancillary airline charge to a particular trip by
recognizing that the flight matches a TMC booking. SLATE also prevents duplicate
information. One example: A traveler may have booked multiple flights in order
to have options but later canceled all but one; SLATE removes the costs of
those canceled flights in order to get at the true cost of the trip.
"The particular problem we have in travel
is: How do we group together transactions for a single trip, especially
multi-legged journeys?" said Choe. "Often, there is no common identifier
for employees across different systems. If they gave themselves an hour, an
experienced traveler could do this, but how do you do it with a million
transactions across a business? If you want to analyze how much it costs to
send people to do business on a particular route, you can't really do it—or
figure out which routes are incurring the highest number of ancillaries. This
technology simply wasn't available five years ago. It can learn the trip
patterns of a particular organization or even of a group within that organization."