Verizon

While this Hive product ran on its own, there was an opportunity to layer in a related solution in a web application w/algorithm to optimize scheduling, automatically receive text transcripts from conference within a monitor of the conference (v. memory & notes), which was delivered to IA centralized database of all co. conferences, effectively creating a high-level searchable database for corporate intelligence.

When developing the sphere video data compression algorithm for Hive to run across basic T1 lines globally, we found an unexpected second-order challenge. For these multi-office multi-nationals, a single meeting could require a couple of days of the work of two to three admins simply to coordinate and reprioritize executive calendars, even using “automated calendaring evites.” 80% of the time, meetings proposed were switched to times based on availability of executives prioritized by seniority versus role or domain. Given the average hourly comp and loss of information and recombinatorial opportunities for brains falling through the scheduling crack, we estimated this “little” issue to be worth in the low millions as far as lost productivity, and without cap on the lost potential for actionable strategy-shifting ideas dependent on the right minds interacting on the right topics at the right times.

So we embedded an intelligence engine in the broader Hive scheduling application that would satisfice based on all schedules, heavier weighing domain expertise across goal/meeting outcome before seniority, then based on second-level conflicts, weighing those conflicts and flexibility based on type of move or reschedule them in the cases where the company potential benefit far outweighed the individual absence. For example, on a next step core infrastructure for data mining meeting, over a 2 week period, the best time was modelled based on the best and second best availabilities for the lead data engineer, lead data strategist, and then VP of data warehousing (most senior, but least involved in these decisions), before moving the CIO from a “personal” (personal training) meeting into this spot, and emailing the contact with a requested move. This was significant enough on its own to warrant development for a stand-alone product, but our priority was in developing a semantic engine that could understand both spoken and body language, pattern meaningful contributions towards company goals, and save core semantic-tagged voice-text archives of any meetings, while making them live searchable across the enterprise. Automatically cross-indexed by category of discussion to create a database across all department conversations on individual topics, and of course, store in long-term archives efficiently for later use in a bigger data map.