Stack AI

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

Links: homepage docs

Provide an enterprise visual platform for building, deploying and governing AI agents on a workflow canvas — orchestrating an AI Agent node over knowledge bases, OpenAPI-described tools, sub-flow tools and multi-project routing, with governance and observability wrapped around every deployment.

Description. Stack AI's documentation describes the product as 'an enterprise platform for building, deploying, and governing AI agents.' Builders 'Build AI workflows with a visual builder' and 'Deploy agents to chat, forms, APIs, and internal teams' while operating 'safely with governance, access controls, and observability.' Core building blocks live in the Core Nodes layer — 'the primary building blocks of every StackAI workflow — the AI reasoning layer, the knowledge retrieval layer, and the platform utility layer.' The AI Agent Node is the central reasoning unit and 'is the heartbeat (or heartbeats!) or your project.' Subflow Tools enable orchestration across a fleet of agents; Knowledge Bases provide RAG; OpenAPI-described tools let the LLM call external APIs.

Agent loop shape. A Stack AI workflow is a graph of Core Nodes — AI Agent (reasoning), Knowledge (retrieval) and platform utility nodes. The AI Agent Node is configured with a provider/model, a system prompt, knowledge sources (indexed Knowledge Base or real-time Search Connected Apps), and tools. Tools are grouped under Providers; each OpenAPI endpoint becomes a distinct tool, and 'the LLM intelligently determines when and how to call these tools based on the context of the conversation and user inputs.' Multi-agent shape is realised through Subflow Tools — 'orchestrations across a fleet of AI agents' that can 'run in parallel and pass results back to the main Agent' — and AI Routing across StackAI Projects. Deployment surfaces include chat, forms, APIs and internal teams.

Primary use cases

enterprise visual building of AI agents with governance and observability
RAG over indexed Knowledge Bases or live Search-Connected Apps
OpenAPI-driven tool integration where each endpoint becomes a callable tool
multi-agent orchestration via Subflow Tools and cross-project routing
deploying agents as chat, forms or APIs to internal teams

flowchart TD USR[User / API / form / chat] --> WF[Workflow canvas] WF --> AGT[AI Agent Node<br/>provider + model + prompt] AGT --> KB[Knowledge Bases — indexed RAG] AGT --> SCA[Search Connected Apps — live search] AGT --> TOOLS[Tools — OpenAPI endpoints + built-ins] AGT --> SUB[Subflow Tools] SUB --> SUB_A[Sub-agent A — parallel] SUB --> SUB_B[Sub-agent B — parallel] SUB_A -.results.-> AGT SUB_B -.results.-> AGT WF --> OUT[Output node — chat / form / API] GOV[(Governance / access controls / observability)] --> WF

Key concepts

AI Agent Node → tool-use (docs) — Core reasoning unit; configures LLM, tools, knowledge sources and agentic behaviour.
Knowledge Bases → agentic-rag (docs) — Indexed content store queried at run time via semantic search; alternative to live Search Connected Apps.
Tools (OpenAPI / built-in) (docs) — Tools grouped under Providers; each OpenAPI endpoint becomes a distinct tool the LLM can call.
Subflow Tools → supervisor (docs) — Sub-workflows attached to an AI Agent as callable tools — used for multi-agent orchestration.
Workflow builder → visual-workflow-graph (docs) — Visual canvas where Core Nodes are wired into a project; outputs deploy to chat, forms, APIs.