Relevance AI

Type: low-code · Vendor: Relevance AI · Language: Web (visual) · License: proprietary (hosted) · Status: active · Status in practice: mature

Links: homepage docs

Provide a hosted, enterprise visual platform on which domain experts assemble individual AI Agents (with tools, knowledge and sub-agents) and compose them into a Workforce — a multi-agent team — gated by approvals, escalations and an oversight dashboard for safe autonomous execution.

Description. Relevance AI is a hosted enterprise agent platform. The homepage frames it as 'The Enterprise Platform for Agents You Can Trust at Scale' where 'your domain experts and AI agents achieve results together, safely, at enterprise scale.' Core building blocks are Agents ('AI entities that autonomously complete tasks like human employees'), Tools ('no-code workflow builder for integrations and automations'), Knowledge ('RAG solution giving agents access to specific information beyond pre-trained knowledge'), and Workforce ('multi-agent teams where specialized AI agents collaborate on complex tasks'). Approvals and escalations are first-class: 'Approvals and escalations are essential components of Workforce that ensure your AI agents operate with appropriate oversight.'

Agent loop shape. An Agent is configured with an LLM core, an instruction set, a set of Tools (no-code workflow tools that wrap integrations), one or more Knowledge bases (RAG), and optionally Sub-agents. The agent autonomously selects tools to call. Sub-agents can be delegated to with Auto-Run for routine tasks or Approval Required for sensitive operations. Workforce is the multi-agent layer: it 'link[s] multiple agents together on a visual canvas to handle complex, multi-step workflows,' adds approvals and escalations as oversight gates, and routes tasks across specialised agents. The 'Workforce Task View serves as your central dashboard for monitoring and managing approval requests.'

Primary use cases

hosted enterprise AI agent building with no-code tools
multi-agent Workforce orchestration across domain-specific agents
human approval workflows for high-risk agent actions
RAG via Knowledge bases with semantic search
scaling repetitive knowledge-work tasks 24/7

flowchart TD USR[User / trigger] --> WF[Workforce — visual canvas] WF --> AGT1[Agent A — LLM + tools + knowledge] WF --> AGT2[Agent B — LLM + tools + knowledge] AGT1 -.delegate.-> SUB[Sub-agent<br/>Auto Run / Approval Required] AGT1 --> TOOLS[Tools — 1000+ connectors / MCP / custom] AGT1 --> KN[(Knowledge — RAG with semantic search)] AGT1 --> APP{Approval gate} APP -- approve --> EXEC[Execute action] APP -- deny --> AGT1 AGT1 -.escalate.-> HUM[(Human reviewer<br/>Workforce Task View)] AGT2 --> APP EXEC --> OUT[Result]

Key concepts

Agent (docs) — Autonomous AI entity with LLM, tools, knowledge and optional sub-agents.
Tool → tool-use (docs) — No-code workflow that wraps an integration or automation and is callable by an agent.
Knowledge → agentic-rag (docs) — RAG layer — knowledge bases with semantic search agents query to ground answers.
Sub-agent (docs) — Agent attached to a parent agent as a delegated task executor; Auto Run or Approval Required mode.
Workforce → supervisor (docs) — Multi-agent team on a visual canvas with approvals, escalations and a task dashboard.
Approval & Escalation → approval-queue (docs) — Oversight gates — agents draft actions and wait for human approval; escalate when uncertain.