New Investment in AI for Voice AI Startup, ‘Podonos’
New Investment in AI for Voice AI Startup, ‘Podonos’
NAVER D2SF, the CVC arm of NAVER, has made a new investment in Podonos, a startup specializing in Voice AI performance evaluation.
The Importance of AI Performance Evaluation
AI models are flooding the global market. Epoch AI, a leading nonprofit AI research organization, currently has over 2,900 AI models registered in its database, with 500–800 actively used in commercial applications. What’s remarkable is the accelerating pace of releases: while 2024 saw new models emerging monthly, 2025 has shifted to a weekly cadence.
How can we compare the performance of so many AI models? Every newly launched AI model claims exceptional benchmark scores. Traditional benchmarking approaches have failed to keep pace with AI advancement, losing their discriminatory power as evaluation standards. We now need innovative AI testing and evaluation frameworks.
The Rise of Voice AI
Voice AI evaluation represents an increasingly critical and valuable market opportunity. Here’s why.
Unlike text-based AI, Voice AI requires multifaceted evaluation criteria. Perfect pronunciation doesn’t guarantee natural intonation. Emotional expression and persona matter significantly, and Voice AI excelling in one language often sounds unnatural in another. This complexity creates substantial entry barriers for companies that successfully earn market trust.
The Voice AI market itself holds tremendous appeal. Since voice represents the most natural form of human communication, the Voice AI market is expanding rapidly and will continue its explosive growth. Voice AI is already proliferating across speech recognition and synthesis, customer service, and audio-video content, serving as the primary interface for rapidly growing AI agents.
However, no single outstanding model can satisfy all market demands. Voice preferences vary dramatically across regions, ages, genders, industries, and contexts. In advertising, the ideal Voice AI must adapt to target user characteristics. Meanwhile, healthcare and finance require verification of pronunciation accuracy for both everyday language and technical terminology to ensure precise information delivery. Consequently, customer demand for Voice AI performance evaluation will inevitably diversify and persist.
Podonos, Unparalleled Evaluation System Combining AI and Human-in-the-Loop
Podonos, backed by NAVER D2SF, delivers Voice AI model performance evaluations tailored to customer needs within 12 hours using a human-in-the-loop (HITL) approach that combines a globally distributed workforce of 150,000 evaluators with proprietary AI automation solutions.
Podonos currently serves both AI model suppliers and consumers, rapidly building market trust across both segments.
Consider AI model developers first. Global AI startups, including Resemble AI, Play AI, and Sanas AI, already recognize Podonos’s value and quality. Companies increasingly publish Podonos evaluation reports instead of traditional benchmarks with significant gaps from real-world performance, using these as evidence of model capabilities. Tech media and communities frequently cite Podonos evaluation results. These performance reports also serve as post-training data for fine-tuning AI models to enhance performance—critical for staying ahead in the fast-moving AI landscape.
From AI model users, it provides a way to explore which model works best for their goals and environment among countless options. Podonos supports customizable performance criteria based on client keywords and requirements, not just average performance across generic standards. Healthcare, finance, gaming, and advertising sectors show particularly active demand for optimal AI model selection, while increasingly sophisticated requirements emerge for hyperpersonalized Voice AI personas tailored to user profiles.
As everyone talks about AI, what market opportunities exist? While many are jumping into AI model competition, Podonos was already thinking one step ahead about the critical questions: how do we evaluate and validate these numerous AI models? How do we find the right AI model for our needs? NAVER D2SF recognizes the growing importance of “AI for AI” as AI advances. We will continue supporting entrepreneurs who view markets from fresh perspectives and explore emerging opportunities.
