Company

MOST LLC is an artificial-intelligence company

We build MOST Insight, a document intelligence platform that lets organisations ask questions of their own files and get cited, permission-aware answers. MOST LLC is a limited liability company registered in Wyoming, United States.

What we do

Companies accumulate enormous archives: policies, contracts, manuals, project plans, spreadsheets and years of e-mail, spread across Microsoft 365, Google Workspace, Dropbox, Box, file servers, laptops and phones. The knowledge is there, but finding it takes minutes or hours — and often the answer is simply re-asked of a colleague.

MOST LLC applies modern AI — retrieval-augmented generation, hybrid vector and keyword search, and large language models — to that archive. The result is MOST Insight: a platform that reads every document a user is permitted to see and answers their question with the sources attached, in the user’s own language.

How we work

We develop and operate the full stack ourselves: connectors, the indexing pipeline, the retrieval engine, the clients and the administrative tooling. Deployments run in our cloud or entirely inside a customer’s network, and are delivered as a containerised platform with automated TLS, health monitoring and cost metering.

Our team works with customers across Europe and North America, from manufacturing groups with terabytes of engineering documents to professional-services firms with sensitive client files.

Principles

What guides our engineering

Truthful by construction

We build AI that shows its evidence. An answer without a source is not an answer.

Access is sacred

The permissions your organisation already defined are the permissions our AI respects, on every channel.

Engineering depth

A Rust indexer, a TypeScript retrieval engine and a vector database tuned for terabyte-scale archives.

Partners, not vendors

Every customer gets an engineer who understands their documents, not a ticket queue.

Technology

Built on a serious stack

Open standards and proven components, assembled and tuned by our own engineers.

Rust indexerParallel crawling, text extraction, chunking and embedding at terabyte scale.
Qdrant vector databaseDense + sparse hybrid retrieval with per-chunk access-control payloads.
Node.js / TypeScript servicesRetrieval orchestration, LLM calls, SSO, admin API and bots.
Large language modelsAzure OpenAI by default; private or open-weight models on request.
Microsoft Graph & Google APIsFirst-class connectors for Microsoft 365 and Google Workspace.
Model Context ProtocolExpose company knowledge to any MCP-capable agent.
Docker & CaddyReproducible deployment with automatic TLS certificates.
Prometheus metricsLatency, token usage and cost per answer, exported for your monitoring.

Talk to the team

Questions about the company, partnerships or press? We answer every message personally.