Ofer Ronen is the Co-Founder and CEO of Tomato.ai, a platform that gives an AI powered voice filter to melt accents for offshore agent voices as they converse, leading to improved CSAT and gross sales metrics.
Ofer beforehand bought three tech startups, two to Google, and one to IAC. He spent the previous 5 years at Google constructing contact middle AI options inside the Space 120 incubator. He closed over $500M in offers for these new options. He holds an MS in Laptop Engineering with a give attention to AI from the College of Michigan, and an MBA from Cornell.
What initially attracted you to machine studying and AI?
AI has had an extended historical past of begins and stops. Durations when there was loads of hope for the know-how to rework industries, adopted by intervals of disillusionment as a result of it didn’t fairly stay as much as the hype.
Once I was doing a Masters in AI a few a long time in the past, on the College of Michigan, it was a interval of disillusionment, when AI was not fairly making an impression. I used to be intrigued by the concept computer systems may very well be taught to carry out duties via examples vs the standard heuristics, which requires fascinated by what express directions to offer. On the time I used to be working at an AI analysis lab on digital brokers which assist lecturers discover sources on-line for his or her lessons. Again then we didn’t have the massive information, the highly effective compute sources, or the superior neural networks that we have now right now, so the capabilities we constructed have been restricted.
From 2016 to 2019 you labored at Google’s Space 120 incubator to design extremely sturdy digital brokers for the most important contact facilities. What was this resolution exactly?
Extra just lately I labored at Google’s Space 120 incubator on among the largest voice digital brokers deployments, together with a few tasks for Fortune 50 corporations with over 100 million assist calls a yr.
In an effort to construct extra sturdy voice digital brokers that may deal with advanced conversations, we took thousands and thousands of historic conversations between people and used these conversations to detect the kind of follow-up questions clients have past their preliminary acknowledged problem. By mining follow-up questions and by mining other ways clients phrase every query, we have been capable of construct versatile digital brokers that may have meandering conversations. This mirrored higher the sort of conversations clients have with human brokers. The tip consequence was a cloth enhance within the whole calls totally dealt with by the digital brokers.
In 2021 and 2022, you constructed a 2nd startup at Space 120,. may you share what this firm was and what you discovered from the expertise?
My second startup inside Space 120 was once more targeted on name facilities. Our resolution targeted on decreasing buyer churn by proactively reaching out to clients proper after a failed assist name the place the shopper expressed their problem however didn’t get to a decision. The outreach can be accomplished by digital brokers skilled to handle these open points. What I discovered from that have is that churn is a troublesome metric to measure in a well timed method. It will possibly take 6 months to get statistically important outcomes for adjustments in churn. That makes it onerous to optimize an expertise quick sufficient and to persuade clients an answer is working.
May you share the genesis story behind your third third contact middle AI startup Tomato.ai, and why you selected to do it your self versus working inside Google?
The concept for Tomato.ai, my third contact middle startup, got here from James Fan, my co-founder and CTO. James thought it might be more practical to promote wine utilizing a French accent, and so what if anybody may very well be made to sound French?
This was the seed of the thought, and from there our considering advanced. As we investigated it extra we discovered a extra acute ache level felt by clients when talking with accented offshore brokers. Clients had issues with comprehension and belief. This represented a bigger market alternative. Given our backgrounds, we realized the sizable impression it might have on name facilities, serving to them enhance their gross sales and assist metrics. We now check with this kind of resolution as Accent Softening.
James and I beforehand led and bought startups, together with every of us promoting a startup to Google.
We determined to depart Google to begin Tomato.ai as a result of, after a few years at Google, we have been itching to get again to beginning and main our personal firm.
Tomato.ai solves an necessary ache level with name facilities, which is softening accents for brokers. May you focus on why voice filters are a most popular resolution to agent coaching?
At Tomato.ai, we perceive the significance of clear communication in name facilities, the place accents can typically create boundaries. As an alternative of relying solely on conventional agent coaching, we have developed voice filters, or what we name “accent softening.” These filters assist brokers preserve their distinctive voice, whereas decreasing their accents, enhancing readability for callers. Through the use of voice filters, we guarantee higher communication and construct belief between brokers and callers, making each interplay more practical and satisfying to the shopper. So, in comparison with in depth coaching packages, voice filters provide an easier and extra rapid resolution to handle accent-related challenges in name facilities.
As present brokers leverage these instruments to boost their efficiency, they are going to be empowered to command increased charges, reflecting their elevated worth in delivering distinctive buyer experiences. Concurrently, the democratizing impact of generative AI will carry new entry-level brokers into the fold, increasing the expertise pool and driving down hourly charges. This dichotomy signifies a elementary transformation within the dynamics of name middle providers, the place know-how and human experience reshape the panorama of the trade, paving the way in which for a extra inclusive and aggressive future.
What are among the totally different machine studying and AI applied sciences which can be used to allow voice filtering?
This kind of real-time voice filtering resolution wouldn’t have been potential simply a few years in the past. Developments in speech analysis mixed with newer architectures just like the transformer mannequin and Deep Neural Networks, and extra highly effective AI {hardware} (like TPUs from Google, and GPUs from NVidia) make it extra potential to construct such options right now. It’s nonetheless a really troublesome downside that requires our crew to invent new methods for coaching speech-to-speech fashions which can be low latency, and top quality.
What sort of suggestions has been acquired from name facilities, and the way has it impacted worker churn charges?
We’ve got sturdy demand from massive and small offshore name facilities to check out our accent softening resolution. These name facilities acknowledge that Tomato.ai may help with their high two issues (1) offshore brokers’ efficiency metrics are less than par vs onshore brokers (2) it’s troublesome to seek out sufficient certified brokers to rent in offshore markets like India and The Philippines.
We count on within the coming weeks to have case research that spotlight the kind of impression name facilities expertise utilizing Accent Softening. We count on gross sales calls to see a right away carry in key metrics like income, shut charges, and lead qualification charges. On the identical time, we count on assist calls to see shorter deal with instances, fewer callbacks, and improved CSAT.
As talked about above churn charges take longer to validate, and so case research with these enhancements will come at a later date.
Tomato.ai just lately raised a $10 million funding spherical, what does this imply for the way forward for the corporate?
As Tomato.ai gears up for its inaugural product launch, the crew stays steadfast in its dedication to reshaping the panorama of world communication and the way forward for work, one dialog at a time.
Thanks for the nice interview, readers who want to be taught extra ought to go to Tomato.ai.
