stordi.com

What It Is

Stordi positions itself as “The Dropbox for AI Agents.” At its core, it provides an API that lets AI Agents store, search, and recall files. According to the site, files, notes, calls, PDFs, images, and other content can be ingested via SDK or MCP. Stordi then handles extraction, chunking, embedding, and indexing, making enterprise files searchable by Agents through semantic queries.

Core Capabilities

Its main focus is the RAG/memory layer for enterprise files, with a clear three-step workflow: Store, Index, and Recall. The demo workspace shows unified querying across PDFs, spreadsheets, screenshots, Markdown notes, PPT files, and legal documents. Users can ask questions such as “Which accounts are at renewal risk?”, “What changed in Q2 revenue?”, or “What legal blockers exist for enterprise procurement?” Responses include source files and supporting snippets, which is important for traceability in enterprise use cases. However, the page does not disclose the specific models used, embedding approach, access control mechanisms, retrieval evaluation metrics, or whether Chinese documents and Chinese-language queries are supported.

Pricing and Availability

Stordi is currently in the early API access stage. The page provides a waitlist entry and says an API key will be sent once access opens. There is no visible information about a free tier, trial policy, official pricing, plan limits, or payment methods. As such, it currently looks more like an early developer preview than a mature commercial SaaS product.

Pros and Cons

The main advantage is its clear positioning: file memory and semantic recall built directly for AI Agents. Its mention of SDK/MCP support also suggests a focus on developer integration. Support for multiple file types and cited sources makes it suitable for enterprise knowledge bases, sales, customer support, legal, and operations analytics. The downside is that many key details are missing, including data privacy, security compliance, Chinese-language support, API documentation, stability, and cost structure. Output quality can only be judged from the demo for now, so its performance in real, complex scenarios remains uncertain.

Who It’s For and Access from China

Stordi is suitable for development teams building enterprise Agents, internal knowledge-base Q&A, cross-document search, or file memory systems. The site does not provide enough information to assess access from mainland China, so its availability there is unknown. Payment methods are also not disclosed. If you need something immediately usable, you may want to consider building your own vector database and RAG pipeline, or choosing an existing enterprise knowledge-base/document Q&A platform instead.