Zoho Goes All in on AI with 25+ Agents, 3 Zia LLMs, and an MCP Server

Its suite includes speech-to-text, text-to-speech, and natural language understanding products that help businesses transcribe, analyze, and monitor multilingual interactions. For many tech companies, the cost-benefit analysis and perceived market opportunity doesn’t justify the investment. Supporting child-specific speech recognition is often viewed as a high-effort, low-return undertaking.

Botlhale AI Brings Speech Recognition to African Languages in Call Centers

Zoho says that the LLM is designed to be privacy-preserving and cost-efficient, with different models tailored for structured and unstructured data. “The 1.3 billion parameter model is not a distilled version of the 7 billion. Children with diverse accents, neurodivergent learners, and multilingual students are disproportionately affected by ASR inaccuracies. These groups are already at a higher risk of being misunderstood by general-purpose models, and when speech AI fails them, it can exacerbate existing disparities in education and healthcare. For AI practitioners, this underscores the need to design systems that are not only accurate but equitable.

We built the AI layer, we built the search layer, we want to build all the tools, all the infrastructure so that we can optimise on cost and accuracy,” Vembu said. Another key challenge is the lack of phoneme-level transcription in many ASR systems. Breaking speech down into individual sounds allows models to track mispronunciations, hesitations, and fluency with far greater precision. This granular approach is especially valuable in educational and therapeutic settings, where understanding subtle differences in speech can inform interventions. Harshal Shah is a voice technology specialist passionate about bridging human expression and machine understanding through inclusive voice solutions. There are even platforms being developed where individuals can contribute their speech patterns, helping to expand public datasets and improve future inclusivity.

Why General-Purpose Speech AI Falls Short for Children

By enhancing articulation, filling in pauses or smoothing out disfluencies, AI acts like a co-pilot in conversation, helping users maintain control while improving intelligibility. This year’s winner is Stephanie Cohen, chief strategy officer at Cloudflare. She is leading efforts to redefine the economic model of the internet, and creating a sustainable future for content creators, publishers, AI companies and the internet at large. “At Algorized we utilize AI for people sensing, so we enable physical AI, we are at the intersection of human and machine,” Lopareva said in her acceptance speech. The award went to her entire team and all the amazing women working at the company. A major infrastructure addition is Zoho’s new MCP server, aimed at enabling interoperability across applications and agents.

• This means collecting diverse training data, supporting non-verbal inputs, and using federated learning to preserve privacy while continuously improving models.
• Zoho has rolled out 25+ AI agents, including pre-built tools like Candidate Screener and Revenue Growth Specialist.
• Supporting users with disabilities is not just ethical, it is a market opportunity.
• So, when a child speaks, these models frequently misinterpret their words or fail to respond entirely.
• Zoho says that the LLM is designed to be privacy-preserving and cost-efficient, with different models tailored for structured and unstructured data.

To truly support children, speech AI must go beyond basic transcription and be purpose-built for the real-world complexities of classrooms, clinics, and other dynamic learning environments. The most effective systems don’t just assign scores or labels; they provide detailed, actionable insights through features like timestamps, phoneme-level transcriptions, and indicators of hesitation. To better understand how inclusive AI speech systems work, let us consider a high-level architecture that begins with nonstandard speech data and leverages transfer learning to fine-tune models. These models are designed specifically for atypical speech patterns, producing both recognized text and even synthetic voice outputs tailored for the user. The company builds speech recognition and analytics tools in African languages – like isiZulu, Sesotho, and Setswana – designed for enterprise call centers.

• Some developers are even integrating facial expression analysis to add more contextual understanding when speech is difficult.
• Explore the future of AI on August 5 in San Francisco—join Block, GSK, and SAP at Autonomous Workforces to discover how enterprises are scaling multi-agent systems with real-world results.
• This award honors a female leader who has helped mentor other women in the field of AI, providing guidance and support and/or encouraging more women to enter the field.
• Many existing tools rely on third-party servers to process speech data – a practice that might suffice for a customer service chatbot but is wholly inappropriate for young learners.
• It currently supports 11 South African languages and is expanding to new markets, including Ghana, Kenya, and Nigeria.

This award honors a woman in the early stage of her AI career who has demonstrated exemplary leadership traits. This award honors a woman who demonstrates exemplary leadership and progress in the emerging field of responsible AI. This award honors a woman who has made a significant impact in AI research, helping accelerate progress either within her organization, as part of academic research or impacting AI generally. The press conference featured key remarks from Ramprakash Ramamoorthy, director of AI research at Zoho and ManageEngine, along with Mani Vembu, CEO of Zoho. “Missed disclosures in non-English calls are a big compliance risk, especially in sectors like insurance and banking,” says CEO Thapelo Nthite.

Building voice AI that listens to everyone: Transfer learning and synthetic speech in action

Today’s automatic speech recognition (ASR) systems are typically trained on data from adult speakers, often English speakers with clear and consistent speech patterns. So, when a child speaks, these models frequently misinterpret their words or fail to respond entirely. When AI fails to understand what a child is saying, it’s a missed opportunity to support learning, flag potential development concerns, or provide timely interventions. Standard speech recognition systems struggle when faced with atypical speech patterns. Whether due to cerebral palsy, ALS, stuttering or vocal trauma, people with speech impairments are often misheard or ignored by current systems. By training models on nonstandard speech data and applying transfer learning techniques, conversational AI systems can begin to understand a wider range of voices.

According to the World Health Organization, more than 1 billion people live with some form of disability. Accessible AI benefits everyone, from aging populations to multilingual users to those temporarily impaired. I have worked on systems where emotional nuance was the last challenge to overcome. For people who rely on assistive technologies, being understood is important, but feeling understood is transformational. We’d like to congratulate all of the women who were nominated to receive a Women in AI Award and to our winners. Thanks to everyone for their nominations and for contributing to the growing awareness of women who are making a significant difference in AI.

If we want the future of conversation to be truly intelligent, it must also be inclusive. Enterprises adopting AI-powered interfaces must consider not only usability, but inclusion. Supporting users with disabilities is not just ethical, it is a market opportunity.

The models are optimised for low-resource environments and are planned to expand to 15 other Indian languages. Botlhale AI, a South African startup founded in 2019, is helping enterprises understand what’s said in the 70% of customer calls that aren’t in English. Children’s data is highly sensitive and must be handled with care and transparent intentions.

Many existing tools rely on third-party servers to process speech data – a practice that might suffice for a customer service chatbot but is wholly inappropriate for young learners. Explore the future of AI on August 5 in San Francisco—join Block, GSK, and SAP at Autonomous Workforces to discover how enterprises are scaling multi-agent systems with real-world results. Additionally, there is a growing interest in explainable AI tools that help users understand how their input is processed. Transparency can build trust, especially among users with disabilities who rely on AI as a communication bridge. This award honors a female leader who has helped mentor other women in the field of AI, providing guidance and support and/or encouraging more women to enter the field. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI.