Dimension scores are derived from public data and fields; weighted into the composite. Reference only.
Riot Haus is an independent AI systems studio founded in 2025 and based in London / Remote. Rather than positioning itself as a general-purpose chatbot company, it focuses on building “retrieval-first” modular AI systems for real production environments. Its core view is that most RAG projects fail not because the model is not powerful enough, but because foundational layers such as retrieval, permissions, security, and failure handling are treated as afterthoughts—leading to hallucinations, noisy context, and uncontrollable outputs.
The website presents System 01 Oku as its first retrieval infrastructure system. It is currently still in an external review phase before public release, with private walkthroughs available by request. Its technical approach includes a combination of vector search and keyword search, RRF fusion, cross-encoder reranking, scope-aware queries, conversation context handling, and context assembly plus streaming for LLM responses. For data sources, it supports Google Drive OAuth + Picker, local files, and browser uploads, along with document processing steps such as OCR, parsing, chunking, and embedding. Its security design includes JWT authentication, encrypted OAuth tokens, tenant isolation, and scoped permissions.
The website does not disclose public pricing, free quotas, trial options, payment methods, or SaaS plans. It appears to be aimed more at enterprises or teams through system design and private walkthrough delivery. Integration details are relatively specific: it mentions Google Drive, local files, browser uploads, the Qdrant vector database, Postgres full-text search, containerized deployment, and environment isolation. However, there is no public API, SDK, webhook, or self-service deployment documentation.
Its main strength is a professional architectural approach that directly addresses key pain points in production-grade RAG: retrieval quality, permission boundaries, verifiable citations, low-confidence handling, and failure recovery. Compared with tools that simply wrap large language models, it places more emphasis on system reliability. The downside is that the public information remains mostly conceptual and architecture-focused, with a lack of real customer cases, performance metrics, SLA details, model selection information, Chinese-language performance, and a complete product experience. Oku has not yet been officially released publicly, which also limits verifiability.
It is better suited to enterprise technical teams that already have document knowledge bases, complex permission requirements, and a need for high-trust Q&A—for example, contract retrieval, internal knowledge bases, Google Drive document Q&A, and multi-tenant RAG systems. It is not very friendly for individual users who simply want to quickly try an AI assistant. The website does not provide information about access from mainland China, and both network availability and payment support are unknown. For local alternatives, users may consider Dify, LlamaIndex, LangChain, Haystack, Qdrant, Weaviate, Azure AI Search, or domestic knowledge base / RAG platforms.
⚠ This review is compiled from public sources and does not constitute a purchase recommendation. Verify all facts on the vendor's official site. Verify on riot.haus official site.
riot.haus is an Unknown AI Apps provider. TG4G tracks its product information, an overall rating of 7.0/10, and a China-accessibility score of Workable. Click "Visit Official Site" to reach riot.haus directly.