The agent assembled a largely complete picture of Maya Research's core offering (emotionally intelligent open-source TTS with Indian language support) by combining the sparse homepage, HuggingFace model cards, and third-party tutorials, but could not retrieve pricing or commercial service details because the site does not publish them—major gaps remain around monetization strategy and hosted API costs despite references to a cloud console.

• ›Steps [1] and [9] (homepage fetches) returned only minimal marketing copy (meta description: 'voice models that sound native') with no substantive feature, pricing, or market positioning details; the actual website content is severely limited, forcing reliance on external sources.
• ›Steps [11] and [27] (HuggingFace model pages) supplied the real substance: model architecture (3B parameters), supported emotions (20+), languages (Veena for Indian languages, Maya1 for English), and deployment options—these were the most content-bearing resources for understanding what the product does.
• ›Steps [14], [21], [23] (maya1.org and tutorial links from search) provided use-case context, real-time performance claims, and zero-shot voice design details, but were discovered via web search, not the main mayaresearch.ai domain.
• ›Step [30] (search result mentioning console.mayaresearch.ai) created expectation of a public pricing or API dashboard, but step [32] fetch failed and no alternative pricing route was discoverable, leaving the commercial/SaaS layer completely opaque.
• ›The site is minimally agent-ready: no dedicated about/pricing/docs pages (all returned 404 in steps [2], [4], [5], [6]), no structured metadata beyond basic meta tags, no API documentation links, and no clear call-to-action for commercial inquiries—the agent had to reconstruct positioning from fragments across HuggingFace, tutorials, and search results rather than finding a single authoritative source.
• ›The gap between open-source free offering and apparent paid cloud service is unresolved; the agent correctly identified this as a major missing piece but could not bridge it within the website itself.