Documentation
AgentRow documentation
How the platform works, from what an agent is to publishing it behind a chat window. The assistant in the corner of this site answers from these pages, so ask it when a page does not answer your question.
Start here
Build your first agentNine steps from a new agent to a published one: instructions, model, tools, knowledge, memory, workflow, test, publish.Core concepts and vocabularyOrganisations, agents, workflows, apps, tools, toolboxes, knowledge bases, memory, conversations, runs and end users.The chat widgetInstall the widget, sign users in, the JavaScript API, events, customisation and conversations in the widget.Knowledge basesCreate a knowledge base, index documents, test retrieval, and understand how hybrid retrieval works inside a run.
Get started
- What is AgentRow?A visual AI agent builder where the graph you draw is the graph the runtime executes, with every step recorded.
- Core concepts and vocabularyOrganisations, agents, workflows, apps, tools, toolboxes, knowledge bases, memory, conversations, runs and end users.
- Build your first agentNine steps from a new agent to a published one: instructions, model, tools, knowledge, memory, workflow, test, publish.
Build agents
- The agent builder tabsWhat each tab does, from Basic info and Instructions through Guardrails, Approval, Schema, Runtime and Publish.
- Models and providersOpenRouter, OpenAI, Anthropic, vLLM and Ollama providers; per-node model, temperature, context limits, streaming, fallbacks and cost limits.
- Workflows: nodes, edges and the default graphThe default agent workflow, the node reference from input to finalize, and how editing and versions work.
- ToolsTool types, what a tool defines, how results reach the model, tool selection with search_tools, and authenticated tools for signed-in users.
- Knowledge basesCreate a knowledge base, index documents, test retrieval, and understand how hybrid retrieval works inside a run.
- MemoryMemory modes and scopes, how memory is read before the answer and written after it, and the approval policy.
Apps and channels
- Apps and end usersApp tabs, how an app run routes to an agent, and how anonymous and signed-in end users are identified.
- The chat widgetInstall the widget, sign users in, the JavaScript API, events, customisation and conversations in the widget.
- Shopify store assistantInstall the AgentRow app in a Shopify store, link it to an app, and put a shopping assistant in the storefront chat window.
Ship and operate
- Publishing and versionsDraft and published agent versions, graph versions, publishing an app, and what end users run.
- Testing, logs and observabilityWhat a run records: steps, edges, generations, tool calls, retrievals, memory activity, conversation storage, analytics and usage.
- Guardrails, approvals and safetyAgent guardrails, the app-level safety check, tool safety levels, approvals, least privilege for signed-in users and visible failures.