How Observe.AI Is Redefining The Future Of Contact Centres

Generative AI (GenAI) has significantly transformed customer experience and service sectors. Companies worldwide are rapidly adopting AI-powered chatbots and virtual agents, leveraging the capabilities of large language models (LLMs) to enhance their interactions with customers.

The appeal is straightforward: AI chatbots can address customer issues 18% more swiftly, achieving a commendable 71% success rate. This efficiency ultimately results in greater customer satisfaction and higher conversion rates.

Nonetheless, as many companies upgrade their customer care operations, concerns regarding potential job losses and declining customer satisfaction persist.

Significantly, there are limitations to AI’s capabilities. In India, for instance, the multitude of languages presents a challenge for LLMs, resulting in many tickets still being routed to human employees.

Consequently, companies recognize the necessity of empowering human agents with advanced AI tools to improve customer support and sales processes.

Bengaluru-based Observe.AI is dedicated to automating complex tasks for human agents, thus revolutionizing customer support approaches.

Founded in 2017 by Swapnil Jain, Akash Singh, and Sharath Keshava Narayana, Observe.AI has enhanced contact center efficiency for over 350 enterprises with its innovative AI solutions.

Jithendra Vepa joined as CTO in 2018 and became a co-founder later, while Singh and Narayana exited in 2022.

The Genesis Of Observe.AI

In an interview with Inc42, Jain, a former technical lead at Twitter (now X), remarked that in 2016, while many were still skeptical about AI, he saw a pivotal moment in the evolution of speech technology and deep learning.

“The rapid advancements in machines’ comprehension of human speech revealed an opportunity for transformative innovations in this domain,” he stated.

During this pivotal time, early voice assistants like Google Assistant, Siri, and Alexa emerged, utilizing natural language processing (NLP) in basic command-based interactions.

To comprehend the technology’s potential, Jain spent six months immersed in a Philippines-based contact center with hundreds of agents, noting how manual processes constrained their capabilities.

Vepa, having worked on speech technology research after his PhD, was also instrumental in the development of text-to-speech systems at Samsung.

Jain met Vepa about a year into the venture, seeking expertise to enhance the core technology of Observe.AI’s quality assurance products for contact centers.

Initially recruited as chief scientist, Vepa quickly ascended to co-founder and CTO.

Confronting challenges across communication channels such as chat and email, the founders prioritized solving the voice interaction issue, leading to their conversational intelligence platform.

Since the 2019 launch of its first product focusing on analytics and manual QA support, Observe.AI has since expanded its offerings, enhancing contact center performance through real-time insights, summaries, and various AI-driven tools.

During the pandemic, a surge in demand for voice technologies reportedly led Observe.AI to grow its business by more than 2X, culminating in acquiring its first significant customer, National Debt Relief, in 2020.

To date, Observe.AI has raised over $214 million from investors, including Nexus Venture Partners, Y Combinator, Scale Venture Partners, and a notable $54 million investment from Menlo Ventures in 2020. It’s also supported by SoftBank.

While the founders did not disclose revenue figures, Observed.AI serves numerous prominent clients, including Pearson, Cox Automotive, Accolade, and DailyPay, primarily in the US’s healthcare, BFSI, travel, and homecare sectors, with a small presence in India.

What’s At The Core Of Observe.AI’s Tech

Observe.AI’s technology aids enterprises through quality assurance, coaching, and agent support, supplying crucial data for process optimization.

The firm asserts that clients have experienced up to 60% efficiency improvements and over a 75% reduction in quality evaluation time.

“We developed our technology based on real-life observations, crafting AI solutions that cater to genuine needs rather than abstract issues,” Jain shared.

Identifying three pivotal areas for improvement in customer service, Observe.AI first sought to enhance compliance and quality analysis of agent-customer conversations to boost satisfaction and sales.

Secondly, they aimed to streamline agents’ ability to extract information from extensive documents, making conversations more efficient. Lastly, they introduced voice agents powered by GenAI and LLMs to enhance service further.

As GenAI was still emerging, Observe.AI built a robust data pipeline and incorporated transformer models early on.

“AI has always been integral to our vision. From day one, we laid the groundwork in model training, technology creation, and data pipelines. As GenAI matured, we adapted promptly,” asserted the CTO.

Their core technology comprises two key elements: speech-to-text capabilities utilizing automatic speech recognition, complemented by LLMs for reasoning and analysis. Their LLM operates with an impressive 40 billion parameters.

Both core models are custom-built in-house, supplemented by domain-specific data while leveraging open-source foundational models.

“The final piece was acquiring a text-to-speech model that delivers human-like quality—this led to our recent acquisition of Dubdub.ai,” Vepa explained. With these technologies combined, we have established a formidable voice AI platform.

Observe.AI acquired Dubdub.ai in March to enhance its text-to-speech technology, now compatible with over 50 languages, including several Indic tongues, and capable of producing audio in diverse voices, emotions, and age groups.

Observe.AI’s Next Leap

While India currently represents a minor segment of Observe.AI’s clientele, Vepa indicated plans for further product developments tailored to this market. However, the country’s diverse linguistic landscape poses a significant obstacle.

Consequently, despite investing heavily in language capabilities, Observe.AI aims to strategically acquire companies proficient in Indic languages, with the Dubdub acquisition serving as a prime example.

Meanwhile, competition intensifies in the conversational AI market, with Indian startups like CoRover and Gnani.ai developing in-house speech recognition models, while larger firms such as NoBroker and Zomato introduce AI agents to improve customer interactions.

Globally, companies like Amazon are also developing similar capabilities. Vepa argues that Observe.AI stands apart as many competitors grapple with latency and accuracy issues, which remain areas of continuous improvement.

“Today’s voicebots often struggle with detecting conversation endpoints. Humans wait to speak until the other party finishes, but machines need to learn this timing,” he noted. “Determining which interactions can be managed by AI is also crucial. Thus, human-AI collaboration is essential for the foreseeable future.”

Vepa further emphasized that emerging players in conversational AI will require time to achieve the trust and consistency necessary for enterprise-grade applications.

The founders perceive a vast market ahead. By 2025, they intend to amplify their voice AI agents, enhancing sales conversions and customer satisfaction through collaborative efforts between humans and AI.

Moreover, they are targeting the acquisition of substantial US firms with high-volume contact center requirements, making it intriguing to observe how Observe.AI’s capabilities will provide an edge over larger players in the conversational AI arena.

[Edited By Shishir Parasher]