> For the complete documentation index, see [llms.txt](https://docs-araistotle.facticity.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs-araistotle.facticity.ai/info-hub/case-study-ai-seer-google-cloud.md). # Case Study: AI Seer × Google Cloud **Industry:** Technology\ **Location:** Singapore ### Overview AI Seer partnered with **Google Cloud** to scale Facticity.AI — a high-accuracy, multi-modal fact-checking platform — making real-time truth verification accessible to a global audience. By leveraging Cloud Run, Artifact Registry, Vertex AI, and Gemini models, AI Seer achieved: * **66.67% higher request throughput** (vs Nov 2024) * **Latency reduction from 14,693 ms to 17 ms** * Faster development cycles and lower operational costs * Significant reduction in AI hallucinations through multi-LLM validation and human feedback loops This collaboration enabled AI Seer to deploy production-grade fact-checking infrastructure while maintaining transparency, explainability, and reliability at scale. ### The Challenge In an environment where misinformation spreads faster than verification, AI Seer needed to: * Scale fact-checking during traffic spikes * Reduce latency without sacrificing accuracy * Minimize hallucinations from generative AI systems * Support rapid experimentation with a lean team Traditional infrastructure and single-model AI approaches were insufficient for production-grade truth verification. *** ### The Solution Using **Google Cloud’s serverless and AI stack**, AI Seer built a scalable and resilient system: * **Cloud Run & Artifact Registry**\ Containerized fact-checking services with auto-scaling and fast rollouts. * **Vertex AI & Gemini**\ Multi-LLM orchestration to cross-validate outputs, reduce hallucinations, and optimize cost-performance. * **Human Feedback Loops**\ Community moderation and expert review dynamically refine model outputs over time. This architecture allowed AI Seer to handle surges in queries while maintaining high accuracy and explainability. *** ### The Impact * **66.67% increase in requests served per second** * **Latency reduced from \~14.7s to 17ms** * Faster feature releases and shorter development cycles * Improved collaboration across technical and non-technical teams * Lower inference and infrastructure costs > “Our biggest concern with adopting generative AI is hallucinations.\ > By using a multi-LLM approach with continuous feedback loops, we significantly reduce hallucinations and improve reliability.”\ > — **Dennis Yap, Founder, AI Seer** ### Why This Matters for ArAIstotle The same infrastructure and methodologies power **ArAIstotle**, including: * Real-time, source-backed verification * Multi-model cross-checking * Community feedback and moderation (Truth-to-Earn) * High-availability performance during viral events This case study demonstrates that ArAIstotle is not an experimental bot — it is built on **enterprise-grade infrastructure** designed for accuracy, scale, and trust. ### Looking Ahead AI Seer continues to build on Google Cloud, exploring: * Gemini 2.0 Flash for faster inference * Expanded browser and document integrations * Deeper multimodal and behavioral analysis Founded in 2019, AI Seer remains committed to becoming the filter for misinformation by embedding authentication directly into information streams. Read full case study: