Relay is organizational memory software for AI-native teams. It captures commitments, decisions, and policies from meetings, threads, and tickets, then feeds that material to the tools already in use. The product presents this as knowledge infrastructure for AI-native teams and centers its workflow on cited, checkable statements.
Its core unit is the atom, described as one sentence and one claim. Each atom includes a speaker, source, timestamp, confidence value, version, and a stable citation ID. The page also shows atom types such as decision, commitment, risk, fact, question, and blocker. Relay can surface answers backed by the atoms they came from, and it includes conflict detection, version supersedence, and a knowledge graph where atoms, entities, and conflicts are queryable.
Relay connects to sources through existing OAuth and webhook surfaces, with no agent installed on endpoints. The listed sources include Slack, Telegram, Google Meet, Teams, Fireflies, Gmail, Outlook, GitHub, Linear, Notion, HubSpot, Zendesk, and Datadog. It also provides Python, TypeScript, and MCP SDKs, along with renderers for LangChain, Semantic Kernel, and Claude Projects. The site says atoms are served to the dashboard, the SDKs, and any tool that speaks MCP or HTTP.
Early access is open, and the first 20 teams get a locked rate for 12 months. The page also states SOC 2 Type 2 and says zero training for user data.
In the RAG, search & retrieval space, Relay takes a focused approach. It focuses on simplifying the management and retrieval of organisational knowledge using AI-powered tools. It is built as an open-source project for developers. Relay is open source under the MIT license. It runs on the web and the command line, and it can be self-hosted.
Relay first shipped in 2026. Key capabilities include query organisational memory, data sync, and knowledge graph integration.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do