Symbl for Startups founder profile: How HUEX Labs is building the drive-thru of the future with voice AI insights
This series takes a closer look at companies participating in the Symbl for Startups program. If you have any questions about applying for entry into the program or want to know more about the company featured in this profile, please contact kira.hunter@symbl.ai.
HUEX Labs is working to solve several glaring problems in the retail and fast-food industries through the use of speech recognition technologies and natural language processing; its voice assistant product can deftly handle customer service-related tasks for on-premise commercial use.
This voice assistant, which essentially amounts to a “digital employee,” also improves productivity, reduces labor costs and maximizes upsell opportunities. On the heels of an unprecedented labor crisis across sectors, it doesn’t take much imagination to gauge the potential impact of this type of support from conversation AI technologies.
Below, CTO Kiran Kadekoppa delves into North America’s aging drive-thru infrastructure, why the timing is ripe for conversation AI systems at retail and how Symbl.ai’s API is helping move the company forward.
Tell me about how the idea for HUEX Labs came about. What market need does it fulfill?
Kiran Kadekoppa: At HUEX Labs, we firmly believe human-to-machine interaction is one way to bring operational efficiencies, automation and retail services into the 21st century. As founders we are restaurant operators, academics and enterprise software experts; we have an intimate understanding of how the labor market is impacting the retail industry.
The retail workforce is moving to gig economy roles, freelancing and part-time roles. This has resulted in retailers reducing hours, shutting down and otherwise incurring damage to their bottom lines.
The pandemic accelerated the need to rely on digital channels and contactless interactions across industries. Our conversation AI-based “digital employee,” which works in a hybrid environment (Edge and Cloud), allows staff on the ground to focus on providing the best guest experience to their customers.
How? Our AI mimics the brand’s best-performing employee and achieves human-level accuracy, leveraging our patent-pending IP that can dynamically adapt to the needs of the brand and domain.
At what point did you discover Symbl.ai’s API, and how did you know it was the right fit for HUEX Labs?
KK: Voice AI is extremely challenging in a retail setting as a result of external noise, legacy technology infrastructure (microphonics) and the diversity of language used (regional language nuances and dialects). We had built our own voice AI pipeline for processing the voice input.
Quickly afterwards we discovered that there is a wealth of data hidden in customer orders and very powerful insights could be extracted from these orders, ranging from customer and employee sentiments, keyword extractions, training opportunities, etc.
Because our focus was about perfecting our purpose-built algorithm for helping retailers and other industries that take orders and interact with customers, we started looking for technology providers who could help us achieve our goals and enable our go-to-market strategy. After examining the provider landscape, we felt comfortable with the offerings of Symbl.ai platform and its tooling.
Can you explain some of the benefits of human-to-human speech analysis as it pertains to drive thru operations?
KK: The drive-thru infrastructure across North America is a mixed bag of technology providers old and new which results in challenges that we do not see speaking to Alexa or Siri at home or in a vehicle. Most of the microphonics systems are over a decade old.
This means we do not have access to clean audio because it is a single microphone input, and approximately 20% of the time the customer and guest services audio interactions overlap, which means the customer is required to repeat the order again.
Additionally, the external noise interference that accompanies the audio interactions makes it an extremely challenging environment for order takers. The repetitive nature of order taking also results in human error ranging from 5–7% of the orders, which directly affects customer satisfaction metrics as well.
A very thorough analysis of these human-to-human interactions and the metrics derived from our analysis helps by creating a feedback loop and enabling the guest services personnel to focus on the hospitality aspect of the drive-thru operations.
How does HUEX Labs compare to other kinds of in-person restaurant ordering software?
KK: The process of taking orders is complex because customers have their own unique preferences and are singular in how they express them when they order food. Most software that is live in the market is suitable more for the “command and branch” style, which is a decision tree that is very typical in phone-based ordering.
The key press or voice in menu such as “number 1 or 4” helps map it to equivalent food in the point of sale system easily. Consumer behavior is highly unpredictable — they start with an order, change it and then may delete it after making all the modifications to start all over again.
The natural language processing and the sophistication needed to achieve this kind of dynamic, as well as transporting all of the audio to the cloud and processing, induces a delay to the in-person interaction.
HUEX Labs has taken a fundamentally different approach to address these challenges with an off-the-shelf, Edge-based solution that requires low compute, low memory footprint and works within the existing audio infrastructure at the location.
Which HUEX Labs features have your customers been most excited about?
KK: Our customers love the intelligent insights that we derive from audio interactions. They use the data to train their staff better, improve their operations, manage inventory and achieve better business outcomes.
Customers feel that they were never aware such data points existed in these interactions. Typically, the corporate brand generates reports of location performance. However, there was a huge time gap between the events that occurred and the findings. This resulted in lost sales opportunities because of inventory management challenges. Our voice analysis helps them better manage their personnel as well as improve upon their hospitality.
How do you see HUEX Labs evolving as the company gains users?
KK: We firmly believe voice or contactless interactions with regard to devices is at a huge inflection point. HUEX Labs plans to provide its services to other industry domains where voice-based interactions could eliminate friction and achieve better business outcomes.
HUEX Labs is working with partners including Symbl.ai to prepare ourselves for this growth and provide differentiated services to stand out in the marketplace and support other businesses’ growth.
This interview has been edited for clarity and brevity.
