As expenditure into consumer analytics tools continues to grow, with research leaning heavily on new technology like artificial intelligence (AI) and machine learning (ML), companies are able to understand their consumers and their behaviour at completely new levels. From minute changes in facial expressions when dealing with sales staff, to heat maps in retail stores indicating which products are catching consumers’ attention, emerging new technology is blowing the lid on what companies thought they understood about consumers.
Robbie AI is one of the companies at the forefront of this technological revolution. Pushing the envelope in terms of facial recognition use cases and capabilities, Robbie AI is evolving how businesses analyse consumer behavior. Their AI technology provides ready-to-use KPIs based on automation of data collection and processing of customer insights, employee performance, marketing KPIs, customer acquisition, loyalty, and product analytics.
Given that this information today comes from manually collected data like market research, mystery shoppers, surveys and other basic methods, Robbie has created software that is entirely based on pure, unfiltered human data, gauging things like emotions, behavior, age, and identity.
To learn more about this exciting new technology, AMEinfo spoke with Karen Marquez, CEO and co-founder of Robbie AI, in an extensive two-part interview. You can find the second part here.
Can you give us a brief summary of how your company was founded and what market problems you’ve been trying to address since then?
Robbie AI was founded in the European Union after a 2-year period of grant research on computer vision technologies to evole and surpass existing technology, specifically facial analysis. This is a very challenging problem in computer vision: facial expression and facial recognition in the wild, with commodity cameras, and with no bias for traditionally less represented groups in datasets: seniors, women, and non-white males.
From a technical point of view, in order to make the technology work commercially, our greatest effort was to collect a large and diverse set of training data to reduce any bias in results, and therefore increase accuracy. Secondly, we had to develop algorithms that protect privacy by de-identifying personal information from any processed image, without storing anything but nodal points in dimensional spaces, unlike traditional face analysis, which uses measurements of the face.
From a market perspective, and since our research interest was the real world, the challenge was to integrate a lot of different classifiers into a single solution – one that is almost plug-and-play. We needed to create a solution that was able to solve a long list of real problems for companies operating in the physical world. The solution needed to accomodate for a lack of access to reliable and real-time data, allow integration with legacy systems, as well as not involving any APIs (so that clients wouldn’t need any technical knowledge) while providing actionable KPIs.
How does your technology work, and how/where do you set up your cameras and sensors?
Robbie AI can plug into existing cameras, and/or work with any commodity camera.
Our technology is exclusively based on video streaming, meaning we don’t have special hardware to process data: no smart cameras with GPUs included, no sensors, and no pre-processing on local servers.
We use very easy IoT infrastructure: IP or PoE cameras (Power over Ethernet, wired) connected to the Robbie Box. The Robbie Box is the gateway: it collects the feed from the camera/s, processes the data in the moment of acquisition and sends results to the cloud (no image is ever stored), for continuous updates of dashboards and alerts.
The gateway can be installed as an app on a computer, or for a larger setup of cameras, a mini PC with an internet connection. All the hardware requirements can be found at any IT hardware store.
As for the position of the cameras, it depends on the needs of the client: it can be simply a camera to monitor entries/exits (to measure traffic and collect insights on the people entering/exiting), or one camera per each section you want to monitor (like in a hotel: front desk, restaurant, areas of a showroom, sections of a store, etc.).
Because we don’t use sensors, smart cameras, or CCTV (which has the problem of being a close camera circuit with footage), we don’t store any frame, photo or video. The gateway acquires the image, and in the same very moment algorithms process the information sent to the cloud, de-identified (and hence not associating the identity of the scanned customer).
What kind of stores and businesses is your technology ideal for, and can you give us examples from your existing clients?
Robbie’s technology has been designed for any company operating in the physical world: hospitality, transportation, brick and mortar retail, healthcare (hospitals), defense, security, and payments. They might be seen as different verticals, but all of them share the same basic needs at the data level: customer identification and verification, customer journey, scores of satisfaction, employee performance and loyalty metrics.
Some of our customers and partners work for national security (United States), hospitality, brands with global reach (headquartered in Spain), travel retail (Switzerland) transportation (Japan), and healthcare (UAE).
What benefits does your facial recognition technology offer businesses over existing customer data collection methods?
Robbie AI is focused on utilizing the 80% of customer data lost in daily interactions in the physical world. Unfortunately, physical companies do not have the same level of data availability as online counterparts, and still use old, expensive, inefficient and inaccurate tools such as:
Beacons, GPS and Bluetooth, which lose 50%- 60% of traffic, position and demographics, and provide very limited customer behavior information. These tools are also very dependent on cross-tracking services, which do not have the consent of users, and therefore pose privacy concerns.
Intensive in human force: Some of the data collecttion methods, such as the use of mystery shoppers, are very human capital intensive, and subjective at the same time.
Surveys and similar methods of collection of feedback are biased, moment-based, and utilize limited respondents (5-15%). Additionally, the data takes a long time to process.
Robbie AI solves all these problems by creating the first integrated analytics system entirely based on facial analysis technology, which integrates customer ID with behavior and demographics analyzed in real time through cloud streaming, combining all of our state of the art classifiers into powerful business KPIs, ready to use.
Robbie AI data is unbiased and real-time, providing companies with real time actionable information for human resources, marketing and operations – all in one place – with the possibility of integrating with the legacy systems of companies to integrate its data with their existing PMSs, CRMs, customer experience platforms, etc.
KPIs are/can be industry benchmarked with anonymized data: per city, region, country, etc.
Dates and data search can both be performed, as well as performance comparison (between locations, countries and regions).
KPIs are updated in real-time, based on live video feeds, every few minutes, and alerts can be sent.