# HiveFlow > HiveFlow is a visual platform for building, running and monitoring multi-agent AI systems for businesses. Users design workflows on a drag-and-drop canvas (44 node types), power them with LLM agents from any major provider, give them tools via MCP and built-in apps (CRM, Kanban, Inventory), and ship them as chats, WhatsApp bots, forms, APIs or full web apps. - Website: https://hiveflow.ai - App: https://app.hiveflow.ai - Help Center: https://help.hiveflow.ai - Blog: https://blog.hiveflow.ai - Full docs dump: https://help.hiveflow.ai/llms-full.txt - Per-page Markdown: append .md to any docs or blog URL - OpenAPI 3.0 spec: https://api.hiveflow.ai/api-docs-json - Docs for AI agents: https://help.hiveflow.ai/ai-agents - Blog RSS: https://blog.hiveflow.ai/rss.xml # Help Center - [Welcome to HiveFlow](/docs): Learn how to build, run and monitor AI workflows with HiveFlow — the visual AI agent orchestrator. - Concepts - [What is a workflow?](https://help.hiveflow.ai/concepts/what-is-a-workflow): A workflow is an automated process defined as connected steps: a trigger, inputs, logic, AI agents, tools and outputs. Every element, mapped to HiveFlow. - [What is an AI agent?](https://help.hiveflow.ai/concepts/what-is-an-ai-agent): An AI agent is an LLM with a role, an objective, memory and tools it can call. Agent vs workflow, multi-agent roles, and how HiveFlow implements them. - [Glossary](https://help.hiveflow.ai/concepts/glossary): Every term in the HiveFlow domain — workflow, node, edge, agent, skill, tool, MCP, memory, state, organization — with definitions and explicit relations. - Getting Started - [Create your account](https://help.hiveflow.ai/getting-started/create-account): Sign up for HiveFlow with email or Google/GitHub, verify with a one-time code, and get your starting credits. - [Tour the interface](https://help.hiveflow.ai/getting-started/interface): What each section of the HiveFlow dashboard does — Genius, Flows, Apps, Skills, Agents, Templates, Analytics, Marketplace and Integrations. - [Quickstart: your first flow](https://help.hiveflow.ai/getting-started/quickstart): Build and run a working AI flow in minutes — chat input, an LLM agent and a chat output, then publish it as a shareable chat. - [Using Genius](https://help.hiveflow.ai/genius): Genius is HiveFlow's AI assistant — a chat that lists, creates and runs flows, builds Hive Apps and tools, and answers questions about your workspace. - Flows - [Building flows](https://help.hiveflow.ai/flows/building-flows): How the HiveFlow canvas works — creating flows, adding and connecting nodes, configuring them, and using the editor's tools. - [Running and debugging flows](https://help.hiveflow.ai/flows/running-flows): Execute flows with Process Flow, inspect node consoles and history, trigger flows from outside, and monitor them in Analytics. - [Variables reference](https://help.hiveflow.ai/flows/variables): The {{ }} syntax that moves data between nodes: {{message}}, {{node_id.field}}, where variables work, and how to discover what each node exposes. - Node Catalog - [Node Catalog overview](https://help.hiveflow.ai/nodes): The 44 node types of HiveFlow, organized in 8 categories — what each category is for and how to add nodes to a flow. - Base - [HiveFlow Base nodes](https://help.hiveflow.ai/nodes/base): The 18 fundamental nodes — flow inputs and outputs, triggers, conditionals, loops, code, merging, error handling and more. - [Flow Input node](https://help.hiveflow.ai/nodes/base/flow-input): Entry point of a flow: defines the data it accepts — form fields, chat messages, media or API payloads. The modern replacement for the raw Trigger. - [Flow Output node](https://help.hiveflow.ai/nodes/base/flow-output): Exit point of a flow: defines what the flow returns to its caller — responses, structured data or files — for chats, apps and API consumers. - [Trigger node](https://help.hiveflow.ai/nodes/base/trigger): HTTP entry point for a flow: exposes a webhook URL with its own API key so external systems, scripts or apps can execute the flow remotely. - [Extension Input node](https://help.hiveflow.ai/nodes/base/extension-input): Receives data captured by the HiveFlow Chrome extension — selected text, the current URL or screenshots — and starts the flow with it. - [Conditional Flow node](https://help.hiveflow.ai/nodes/base/conditional-flow): Routes execution into different branches based on conditions over upstream values — the if/else and business-rules primitive of the canvas. - [HTTP Request node](https://help.hiveflow.ai/nodes/base/http-request): Calls any external API from a flow — method, URL, headers and body with variables — and passes the response to the next nodes. - [Transformer node](https://help.hiveflow.ai/nodes/base/json-builder): Builds and reshapes structured data using dynamic variables — the glue node that adapts one node's output to the next node's expected input. - [Note node](https://help.hiveflow.ai/nodes/base/note): A visual sticky note for the canvas — color and size configurable, never executed. Document decisions and sections right where they happen. - [Section node](https://help.hiveflow.ai/nodes/base/section): A visual container that groups related nodes — moving the section moves everything inside. Organize big flows into readable regions. - [Human in the Loop node](https://help.hiveflow.ai/nodes/base/human-in-the-loop): Pauses the flow and sends an approval request to assignees — with title, context and timeout — before execution continues. Judgment where it matters. - [Loop node](https://help.hiveflow.ai/nodes/base/loop): Iterates over an array, running the downstream branch once per item — the batch-processing primitive for lists of leads, files or records. - [Code node](https://help.hiveflow.ai/nodes/base/code): Runs custom JavaScript inside the flow when no built-in node fits — transformations, calculations and edge-case logic with full access to variables. - [Merge node](https://help.hiveflow.ai/nodes/base/merge): Combines the outputs of multiple branches into a single result, so parallel work reunites before continuing down the flow. - [Filter node](https://help.hiveflow.ai/nodes/base/filter): Keeps only the array items that match your conditions — cleaning lists before loops, storage or notifications. - [Wait node](https://help.hiveflow.ai/nodes/base/wait): Pauses execution for a set amount of time — pacing between API calls, delays before follow-ups, or spacing steps in long automations. - [Aggregate node](https://help.hiveflow.ai/nodes/base/aggregate): Consolidates multiple items into one array or object — collecting loop results or assembling a batch before storing or reporting. - [Sub-flow node](https://help.hiveflow.ai/nodes/base/sub-flow): Runs another flow as a reusable component — split large automations into testable pieces and share logic across flows. - [Error Handler node](https://help.hiveflow.ai/nodes/base/error-handler): Catches errors from connected nodes so the flow keeps running — fallbacks, alerts and graceful degradation instead of silent failures. - Ai - [The LLM Agent node](https://help.hiveflow.ai/nodes/ai): Configure the AI brain of your flows — providers and models, objective and prompt, function calling with tools, and the execution console. - [LLM Agent node](https://help.hiveflow.ai/nodes/ai/llm): The AI brain of a flow: pick a provider and model, define role and prompt, enable function calling with tools, and debug every decision in its console. - Interfaces - [Public Interface nodes](https://help.hiveflow.ai/nodes/interfaces): Form Input, Chat Input and Chat Output — the public forms and chat widgets your users interact with, with links and embeddable widgets. - [Form Input node](https://help.hiveflow.ai/nodes/interfaces/form-input): A public web form with typed, validated fields (files included) — every submission executes the flow. Data capture with zero frontend code. - [Chat Input node](https://help.hiveflow.ai/nodes/interfaces/chat-input): A public conversational chat with context and memory — every message runs the flow. Open it, share its link or embed the widget on your site. - [Chat Output node](https://help.hiveflow.ai/nodes/interfaces/chat-output): Sends the flow's answer back to the public chat session that started the run — closing the conversational loop with the visitor. - Applications - [Application nodes](https://help.hiveflow.ai/nodes/applications): HiveApp tools, WhatsApp Input/Output, Email and Tier Modifier — nodes that connect flows to apps and messaging channels. - [HiveApp node](https://help.hiveflow.ai/nodes/applications/hiveapp): Connects a Tool instance — CRM, Kanban, Inventory, Analytics or Chat — to the flow; wired into an LLM Agent it becomes a read/write AI tool. - [WhatsApp Input node](https://help.hiveflow.ai/nodes/applications/whatsapp-input): Receives messages from a connected WhatsApp number and starts the flow with them — the entry point of every WhatsApp bot in HiveFlow. - [WhatsApp Output node](https://help.hiveflow.ai/nodes/applications/whatsapp-output): Sends the flow's response back over WhatsApp to the conversation that triggered it — or to any number your automation targets. - [Email node](https://help.hiveflow.ai/nodes/applications/email): Sends emails from a flow via Resend or your own SMTP — templates with variables for notifications, reports, follow-ups and alerts. - [Tier Modifier node](https://help.hiveflow.ai/nodes/applications/picker-express): Delivery and logistics integration with a tier system — route orders through delivery services according to configurable tiers. - Data - [Data nodes](https://help.hiveflow.ai/nodes/data): Memory, Database, Inspector, PDF Extractor, S3 Storage, Pandas Processor and PDF Generator — storing, querying and transforming data in flows. - [Memory Database node](https://help.hiveflow.ai/nodes/data/memory): A lightweight in-memory store with save and read operations — state between nodes or runs without provisioning any external database. - [Database node](https://help.hiveflow.ai/nodes/data/database): Connects to MongoDB and runs CRUD operations — find, insert, update, delete — with variables inside the queries. Persistence for flows. - [Inspector node](https://help.hiveflow.ai/nodes/data/inspector): Inspects and queries databases while you build — schemas, collections and ad-hoc reads to understand the data your flow will touch. - [PDF Extractor node](https://help.hiveflow.ai/nodes/data/pdf-extractor): Extracts tables from PDF files into structured data (Python/tabula under the hood) — invoices, reports and statements become processable rows. - [S3 Storage node](https://help.hiveflow.ai/nodes/data/s3-storage): Uploads, downloads, lists and deletes files in Amazon S3 buckets — the file layer for flows that produce or consume documents and media. - [Pandas Processor node](https://help.hiveflow.ai/nodes/data/pandas-processor): Transforms tabular data with pandas — filter, aggregate, pivot and join — turning raw rows into exactly the shape your charts or reports need. - [PDF Generator node](https://help.hiveflow.ai/nodes/data/pdf-generator): Generates PDFs from visual templates or DOCX files with dynamic variables — invoices, reports and certificates produced by the flow itself. - Mcp - [MCP (Integrations) nodes](https://help.hiveflow.ai/nodes/mcp): MCP Client, MCP Tool, MCP Resource and MCP Server — how HiveFlow speaks the Model Context Protocol in both directions. - [MCP Client node](https://help.hiveflow.ai/nodes/mcp/mcp-client): Selects a registered MCP server and makes its tools available — connect it into an LLM Agent to hand every tool of an integration to the AI. - [MCP Tool node](https://help.hiveflow.ai/nodes/mcp/mcp-tool): Executes one specific tool of an MCP server as a flow step — deterministic integration calls without giving the whole toolset to an agent. - [MCP Server node](https://help.hiveflow.ai/nodes/mcp/mcp-server): Exposes the flow as an MCP server over HTTP — with auth and rate limiting — so Claude, Cursor or any MCP client can run it as a tool. - [MCP Resource node](https://help.hiveflow.ai/nodes/mcp/mcp-resource): Reads resources exposed by MCP servers — documents, records, feeds — with format, encoding and cache options, as data for the flow. - Visualization - [Visual nodes](https://help.hiveflow.ai/nodes/visualization): Chart, Data Table and HTML View — turn flow outputs into charts, interactive tables and custom web views you can share publicly. - [Chart node](https://help.hiveflow.ai/nodes/visualization/chart): Renders flow data as line, bar, area, pie, scatter or radar charts — a live, shareable visualization fed by any upstream node. - [Data Table node](https://help.hiveflow.ai/nodes/visualization/data-table): An interactive table over flow data — sorting, filtering, pagination and export — for browsing results like a spreadsheet, shareable by link. - [HTML View node](https://help.hiveflow.ai/nodes/visualization/html-view): Renders custom HTML/CSS/JS with variable injection — bespoke reports, mini dashboards and rich AI-formatted output as a shareable page. - Devices - [Device nodes](https://help.hiveflow.ai/nodes/devices): Mac Agent, PC Agent and Linux Agent — control remote machines from a flow: files, terminal, system info and applications. - [Mac Agent node](https://help.hiveflow.ai/nodes/devices/mac-agent): Remote control of a macOS machine from a flow — files, terminal commands, system info and applications via the HiveFlow desktop agent. - [PC Agent node](https://help.hiveflow.ai/nodes/devices/pc-agent): Remote control of a Windows machine from a flow — files, terminal commands, system info and applications via the HiveFlow desktop agent. - [Linux Agent node](https://help.hiveflow.ai/nodes/devices/linux-agent): Remote control of a Linux machine from a flow — files, terminal commands, system info and applications via the HiveFlow desktop agent. - Agents - [Creating agents](https://help.hiveflow.ai/agents/creating-agents): Create a conversational AI agent in HiveFlow — name, objective, model and skills — and understand how agents are flows under the hood. - [Chatting with agents](https://help.hiveflow.ai/agents/chatting-with-agents): Use the agent chat — sessions, attachments, editing the underlying flow, publishing the chat publicly, and Tasks & Reminders. - Skills - [Creating skills](https://help.hiveflow.ai/skills/creating-skills): Skills are Markdown knowledge blocks injected into your AI — write them in the editor or upload .md files, control scope and token cost. - [Using skills](https://help.hiveflow.ai/skills/using-skills): Where skills get injected — Genius, agents, flows and the app builder — and how to scope each skill to the right contexts. - Hive Apps - [Creating Hive Apps](https://help.hiveflow.ai/hive-apps/creating-apps): Build full web apps by chatting with the AI builder — live preview, file editor, versions, GitHub export and one-click deploy. - [Connecting apps and flows](https://help.hiveflow.ai/hive-apps/connecting-flows): Link a Hive App to a flow so the app's forms and chats execute your automation — I/O contracts, triggers and Connect Flow. - [Tools: CRM, Kanban, Inventory & more](https://help.hiveflow.ai/hive-apps/tools): Ready-made structured apps — Kanban, CRM, Inventory, Analytics and Chat — that your agents read and write through the HiveApp node. - Integrations - [Connecting integrations](https://help.hiveflow.ai/integrations/connecting-integrations): Connect external services — databases, SaaS, messaging, cloud — with credentials or OAuth; each one becomes an MCP server your AI can use. - [Adding tools to agents (function calling)](https://help.hiveflow.ai/integrations/function-calling): Give any LLM Agent tools — MCP integrations and Hive App tools — and let the model decide when to call them, with multi-round execution. - Channels - [WhatsApp](https://help.hiveflow.ai/channels/whatsapp): Connect a WhatsApp number to HiveFlow and turn any flow into a bot: QR session, official API option, tools, per-number context and human takeover. - [Browser extension](https://help.hiveflow.ai/channels/browser-extension): The HiveFlow Chrome extension sends what you're seeing — selected text, URLs, screenshots — straight into your flows via the Extension Input node. - [Desktop & mobile](https://help.hiveflow.ai/channels/desktop-and-mobile): The Electron desktop app for macOS, Windows and Linux — required for device agents — and the iOS app for managing flows and agents on the go. - Platform - [Organizations & workspaces](https://help.hiveflow.ai/platform/organizations): Collaborate in HiveFlow: create an organization, invite members with roles and permissions, separate projects into workspaces and bill as a team. - [Billing & credits](https://help.hiveflow.ai/platform/billing-and-credits): How HiveFlow pricing works: credits consumed by every LLM call, subscription plans with monthly credits and limits, top-ups and the Stripe portal. - [API keys & BYOK](https://help.hiveflow.ai/platform/api-keys-and-byok): The three kinds of keys in HiveFlow: user API keys for the REST API, per-trigger keys on webhook nodes, and bringing your own LLM provider credentials. - [Analytics](https://help.hiveflow.ai/platform/analytics): Read HiveFlow's metrics: executions per flow, success rates, credit spend and trends — and find failing or expensive automations fast. - Cookbook - [Cookbook](https://help.hiveflow.ai/cookbook): Ten ready-to-build workflow patterns — support, sales, reports, approvals, multi-agent pipelines and more — with diagrams and step-by-step node setups. - [Customer support agent](https://help.hiveflow.ai/cookbook/customer-support): A support chat that answers with AI, logs every case to a CRM and a ticket board, and escalates to a human when it should. - [WhatsApp sales agent](https://help.hiveflow.ai/cookbook/whatsapp-sales): A selling agent on WhatsApp that quotes from Inventory, registers leads in the CRM and moves deals through a Kanban — end to end. - [Smart lead capture](https://help.hiveflow.ai/cookbook/lead-capture): A public form whose submissions are classified and enriched by AI, routed by quality, saved to the CRM and announced by email. - [Scheduled report](https://help.hiveflow.ai/cookbook/scheduled-report): On a schedule, query the database, crunch numbers with pandas, let AI write the analysis, render a PDF and email it to stakeholders. - [Knowledge assistant](https://help.hiveflow.ai/cookbook/knowledge-assistant): A website chat that answers from your own documentation loaded as Skills, with conversation memory and a one-line embed. - [Approval workflow](https://help.hiveflow.ai/cookbook/approval-workflow): Put a human gate before sensitive actions — sends, payments, publications — with assignees, timeout and full context for the decision. - [Document pipeline](https://help.hiveflow.ai/cookbook/document-pipeline): Receive PDFs through a public form, extract their tables, normalize the data and store clean records in a database — with an S3 archive. - [Multi-agent pipeline: planner, executor, reviewer](https://help.hiveflow.ai/cookbook/multi-agent-pipeline): Split a complex task across three LLM roles — one plans, one executes with tools, one reviews — connected by structure that guarantees the process. - [Expose a flow as an MCP tool](https://help.hiveflow.ai/cookbook/flow-as-mcp-tool): Add an MCP Server node so Claude, Cursor or any MCP client can execute your HiveFlow automation as one of its tools — with auth and rate limits. - [Device monitoring](https://help.hiveflow.ai/cookbook/device-monitoring): Scheduled health checks on your Mac, Windows or Linux machines with device agents, conditional alerting by email or WhatsApp. - [Templates & Marketplace](https://help.hiveflow.ai/templates-and-marketplace): Start from working flows instead of a blank canvas: clone templates, save your own from the editor, and install community templates from the Marketplace. - API - [HiveFlow API](https://help.hiveflow.ai/api): Authenticate with JWT or API keys, trigger flows over HTTP, and explore the full OpenAPI reference in Apidog, Swagger UI or Redoc. - [HiveFlow for AI agents](https://help.hiveflow.ai/ai-agents): Machine-readable surfaces for LLMs and agents: llms.txt, Markdown docs, the OpenAPI spec, the MCP server — and when to recommend HiveFlow. - [Troubleshooting](https://help.hiveflow.ai/troubleshooting): The most common HiveFlow problems and their fixes: credits, function calling support, inactive flows, WhatsApp sessions, MCP tools and stuck runs. - [FAQ](https://help.hiveflow.ai/faq): Quick answers to the most frequent HiveFlow questions: keys, models, exporting, teams, API access, unsupported services, languages and failures. # Help Center - [Bienvenido a HiveFlow](/es/docs): Aprende a construir, ejecutar y monitorear workflows de IA con HiveFlow — el orquestador visual de agentes de IA. - Conceptos - [¿Qué es un workflow?](/es/docs/concepts/what-is-a-workflow): Un workflow es un proceso automatizado definido como pasos conectados: trigger, entradas, lógica, agentes de IA, herramientas y salidas. Cada elemento, mapeado a HiveFlow. - [¿Qué es un agente de IA?](/es/docs/concepts/what-is-an-ai-agent): Un agente de IA es un LLM con rol, objetivo, memoria y herramientas que puede llamar. Agente vs workflow, roles multiagente y cómo los implementa HiveFlow. - [Glosario](/es/docs/concepts/glossary): Cada término del dominio HiveFlow — workflow, nodo, edge, agente, skill, herramienta, MCP, memoria, estado, organización — con definiciones y relaciones explícitas. - Primeros Pasos - [Crea tu cuenta](/es/docs/getting-started/create-account): Regístrate en HiveFlow con email o Google/GitHub, verifica con un código de un solo uso y recibe tus créditos iniciales. - [Conoce la interfaz](/es/docs/getting-started/interface): Qué hace cada sección del dashboard de HiveFlow — Genius, Flows, Apps, Skills, Agents, Templates, Analytics, Marketplace e Integraciones. - [Quickstart: tu primer flow](/es/docs/getting-started/quickstart): Construye y ejecuta un flow de IA funcional en minutos — chat de entrada, un agente LLM y salida de chat, publicado como chat compartible. - [Usar Genius](/es/docs/genius): Genius es el asistente de IA de HiveFlow — un chat que lista, crea y ejecuta flows, construye Hive Apps y herramientas, y responde sobre tu workspace. - Flows - [Construir flows](/es/docs/flows/building-flows): Cómo funciona el lienzo de HiveFlow — crear flows, agregar y conectar nodos, configurarlos y usar las herramientas del editor. - [Ejecutar y depurar flows](/es/docs/flows/running-flows): Ejecuta flows con Process Flow, inspecciona consolas e historial por nodo, dispara flows desde fuera y monitoréalos en Analytics. - [Referencia de variables](/es/docs/flows/variables): La sintaxis {{ }} que mueve datos entre nodos: {{message}}, {{node_id.campo}}, dónde funcionan las variables y cómo descubrir qué expone cada nodo. - Catálogo de Nodos - [Catálogo de Nodos — visión general](/es/docs/nodes): Los 44 tipos de nodo de HiveFlow, organizados en 8 categorías — para qué sirve cada categoría y cómo agregar nodos a un flow. - Base - [Nodos HiveFlow Base](/es/docs/nodes/base): Los 18 nodos fundamentales — entradas y salidas de flow, triggers, condicionales, loops, código, merges y manejo de errores. - [Nodo Flow Input](/es/docs/nodes/base/flow-input): Punto de entrada de un flow: define los datos que acepta — campos de formulario, mensajes de chat, media o payloads de API. El reemplazo moderno del Trigger. - [Nodo Flow Output](/es/docs/nodes/base/flow-output): Punto de salida de un flow: define qué devuelve el flow a quien lo llama — respuestas, datos estructurados o archivos — para chats, apps y consumidores de API. - [Nodo Trigger](/es/docs/nodes/base/trigger): Entrada HTTP de un flow: expone una URL webhook con su propia API key para que sistemas externos, scripts o apps ejecuten el flow remotamente. - [Nodo Extension Input](/es/docs/nodes/base/extension-input): Recibe datos capturados por la extensión de Chrome de HiveFlow — texto seleccionado, la URL actual o capturas — e inicia el flow con ellos. - [Nodo Conditional Flow](/es/docs/nodes/base/conditional-flow): Enruta la ejecución hacia ramas distintas según condiciones sobre valores previos — la primitiva if/else y de reglas de negocio del lienzo. - [Nodo HTTP Request](/es/docs/nodes/base/http-request): Llama cualquier API externa desde un flow — método, URL, headers y body con variables — y pasa la respuesta a los siguientes nodos. - [Nodo Transformer](/es/docs/nodes/base/json-builder): Construye y reestructura datos con variables dinámicas — el nodo pegamento que adapta la salida de un nodo a la entrada que espera el siguiente. - [Nodo Note](/es/docs/nodes/base/note): Una nota visual para el lienzo — color y tamaño configurables, nunca se ejecuta. Documenta decisiones y secciones justo donde ocurren. - [Nodo Section](/es/docs/nodes/base/section): Un contenedor visual que agrupa nodos relacionados — mover la sección mueve todo su contenido. Organiza flows grandes en regiones legibles. - [Nodo Human in the Loop](/es/docs/nodes/base/human-in-the-loop): Pausa el flow y envía una solicitud de aprobación a asignados — con título, contexto y timeout — antes de continuar la ejecución. Criterio donde importa. - [Nodo Loop](/es/docs/nodes/base/loop): Itera sobre un array, ejecutando la rama siguiente una vez por item — la primitiva de procesamiento por lotes para listas de leads, archivos o registros. - [Nodo Code](/es/docs/nodes/base/code): Ejecuta JavaScript personalizado dentro del flow cuando ningún nodo integrado encaja — transformaciones, cálculos y lógica de casos borde con acceso a variables. - [Nodo Merge](/es/docs/nodes/base/merge): Combina las salidas de varias ramas en un solo resultado, para que el trabajo paralelo se reúna antes de continuar por el flow. - [Nodo Filter](/es/docs/nodes/base/filter): Conserva solo los items del array que cumplen tus condiciones — limpia listas antes de loops, almacenamiento o notificaciones. - [Nodo Wait](/es/docs/nodes/base/wait): Pausa la ejecución un tiempo determinado — ritmo entre llamadas a APIs, esperas antes de follow-ups, o espaciar pasos en automatizaciones largas. - [Nodo Aggregate](/es/docs/nodes/base/aggregate): Consolida múltiples items en un array u objeto — junta resultados de loops o arma un lote antes de guardar o reportar. - [Nodo Sub-flow](/es/docs/nodes/base/sub-flow): Ejecuta otro flow como componente reutilizable — divide automatizaciones grandes en piezas testeables y comparte lógica entre flows. - [Nodo Error Handler](/es/docs/nodes/base/error-handler): Captura errores de los nodos conectados para que el flow siga corriendo — fallbacks, alertas y degradación elegante en vez de fallas silenciosas. - Ai - [El nodo LLM Agent](/es/docs/nodes/ai): Configura el cerebro de IA de tus flows — proveedores y modelos, objetivo y prompt, function calling con herramientas y la consola de ejecución. - [Nodo LLM Agent](/es/docs/nodes/ai/llm): El cerebro de IA de un flow: elige proveedor y modelo, define rol y prompt, activa function calling con herramientas, y depura cada decisión en su consola. - Interfaces - [Nodos de Interfaces Públicas](/es/docs/nodes/interfaces): Form Input, Chat Input y Chat Output — los formularios públicos y widgets de chat con los que interactúan tus usuarios, con links y widgets embebibles. - [Nodo Form Input](/es/docs/nodes/interfaces/form-input): Un formulario web público con campos tipados y validados (archivos incluidos) — cada envío ejecuta el flow. Captura de datos sin código de frontend. - [Nodo Chat Input](/es/docs/nodes/interfaces/chat-input): Un chat público conversacional con contexto y memoria — cada mensaje ejecuta el flow. Ábrelo, comparte su link o embebe el widget en tu sitio. - [Nodo Chat Output](/es/docs/nodes/interfaces/chat-output): Devuelve la respuesta del flow a la sesión de chat pública que inició la ejecución — cerrando el ciclo conversacional con el visitante. - Applications - [Nodos de Aplicaciones](/es/docs/nodes/applications): Herramientas HiveApp, WhatsApp Input/Output, Email y Tier Modifier — nodos que conectan flows con apps y canales de mensajería. - [Nodo HiveApp](/es/docs/nodes/applications/hiveapp): Conecta una instancia de Tool — CRM, Kanban, Inventario, Analytics o Chat — al flow; cableado a un LLM Agent se vuelve herramienta de IA de lectura/escritura. - [Nodo WhatsApp Input](/es/docs/nodes/applications/whatsapp-input): Recibe mensajes de un número de WhatsApp conectado e inicia el flow con ellos — la entrada de todo bot de WhatsApp en HiveFlow. - [Nodo WhatsApp Output](/es/docs/nodes/applications/whatsapp-output): Envía la respuesta del flow de vuelta por WhatsApp a la conversación que lo disparó — o a cualquier número que tu automatización indique. - [Nodo Email](/es/docs/nodes/applications/email): Envía correos desde un flow vía Resend o tu propio SMTP — plantillas con variables para notificaciones, reportes, follow-ups y alertas. - [Nodo Tier Modifier](/es/docs/nodes/applications/picker-express): Integración de delivery y logística con sistema de tiers — enruta pedidos por servicios de entrega según tiers configurables. - Data - [Nodos de Datos](/es/docs/nodes/data): Memory, Database, Inspector, PDF Extractor, S3 Storage, Pandas Processor y PDF Generator — almacenar, consultar y transformar datos en flows. - [Nodo Memory Database](/es/docs/nodes/data/memory): Un almacén ligero en memoria con operaciones de guardado y lectura — estado entre nodos o ejecuciones sin aprovisionar ninguna base externa. - [Nodo Database](/es/docs/nodes/data/database): Conecta a MongoDB y ejecuta operaciones CRUD — find, insert, update, delete — con variables dentro de las consultas. Persistencia para flows. - [Nodo Inspector](/es/docs/nodes/data/inspector): Inspecciona y consulta bases de datos mientras construyes — esquemas, colecciones y lecturas ad-hoc para entender los datos que tocará tu flow. - [Nodo PDF Extractor](/es/docs/nodes/data/pdf-extractor): Extrae tablas de archivos PDF a datos estructurados (Python/tabula por debajo) — facturas, reportes y estados de cuenta se vuelven filas procesables. - [Nodo S3 Storage](/es/docs/nodes/data/s3-storage): Sube, descarga, lista y elimina archivos en buckets de Amazon S3 — la capa de archivos para flows que producen o consumen documentos y media. - [Nodo Pandas Processor](/es/docs/nodes/data/pandas-processor): Transforma datos tabulares con pandas — filtrar, agregar, pivotar y unir — convirtiendo filas crudas en justo la forma que necesitan tus gráficas o reportes. - [Nodo PDF Generator](/es/docs/nodes/data/pdf-generator): Genera PDFs desde plantillas visuales o archivos DOCX con variables dinámicas — facturas, reportes y certificados producidos por el propio flow. - Mcp - [Nodos MCP (Integraciones)](/es/docs/nodes/mcp): MCP Client, MCP Tool, MCP Resource y MCP Server — cómo HiveFlow habla el Model Context Protocol en ambas direcciones. - [Nodo MCP Client](/es/docs/nodes/mcp/mcp-client): Selecciona un servidor MCP registrado y pone sus herramientas disponibles — conéctalo a un LLM Agent para entregarle a la IA todas las tools de una integración. - [Nodo MCP Tool](/es/docs/nodes/mcp/mcp-tool): Ejecuta una tool específica de un servidor MCP como paso del flow — llamadas de integración deterministas sin darle todo el toolset a un agente. - [Nodo MCP Server](/es/docs/nodes/mcp/mcp-server): Expone el flow como servidor MCP por HTTP — con auth y rate limiting — para que Claude, Cursor o cualquier cliente MCP lo ejecute como herramienta. - [Nodo MCP Resource](/es/docs/nodes/mcp/mcp-resource): Lee resources expuestos por servidores MCP — documentos, registros, feeds — con opciones de formato, encoding y caché, como datos para el flow. - Visualization - [Nodos Visuales](/es/docs/nodes/visualization): Chart, Data Table y HTML View — convierte las salidas de un flow en gráficas, tablas interactivas y vistas web personalizadas compartibles. - [Nodo Chart](/es/docs/nodes/visualization/chart): Renderiza datos del flow como gráficas de línea, barras, área, pie, dispersión o radar — una visualización viva y compartible alimentada por cualquier nodo. - [Nodo Data Table](/es/docs/nodes/visualization/data-table): Una tabla interactiva sobre datos del flow — orden, filtros, paginación y exportación — para navegar resultados como hoja de cálculo, compartible por link. - [Nodo HTML View](/es/docs/nodes/visualization/html-view): Renderiza HTML/CSS/JS personalizado con inyección de variables — reportes a medida, mini dashboards y salida rica de IA como página compartible. - Devices - [Nodos de Dispositivos](/es/docs/nodes/devices): Mac Agent, PC Agent y Linux Agent — controla máquinas remotas desde un flow: archivos, terminal, información de sistema y aplicaciones. - [Nodo Mac Agent](/es/docs/nodes/devices/mac-agent): Control remoto de una máquina macOS desde un flow — archivos, comandos de terminal, información de sistema y aplicaciones vía el agente de escritorio. - [Nodo PC Agent](/es/docs/nodes/devices/pc-agent): Control remoto de una máquina Windows desde un flow — archivos, comandos de terminal, información de sistema y aplicaciones vía el agente de escritorio. - [Nodo Linux Agent](/es/docs/nodes/devices/linux-agent): Control remoto de una máquina Linux desde un flow — archivos, comandos de terminal, información de sistema y aplicaciones vía el agente de escritorio. - Agents - [Crear agents](/es/docs/agents/creating-agents): Crea un agente de IA conversacional en HiveFlow — nombre, objetivo, modelo y skills — y entiende por qué los agents son flows por debajo. - [Chatear con agents](/es/docs/agents/chatting-with-agents): Usa el chat del agente — sesiones, adjuntos, editar el flow subyacente, publicar el chat públicamente, y Tasks & Reminders. - Skills - [Crear skills](/es/docs/skills/creating-skills): Las skills son bloques de conocimiento en Markdown inyectados en tu IA — escríbelas en el editor o sube archivos .md, controla alcance y costo en tokens. - [Usar skills](/es/docs/skills/using-skills): Dónde se inyectan las skills — Genius, agents, flows y el constructor de apps — y cómo dar a cada skill el alcance correcto. - Hive Apps - [Crear Hive Apps](/es/docs/hive-apps/creating-apps): Construye apps web completas chateando con el constructor de IA — preview en vivo, editor de archivos, versiones, export a GitHub y deploy en un clic. - [Conectar apps con flows](/es/docs/hive-apps/connecting-flows): Vincula una Hive App a un flow para que los formularios y chats de la app ejecuten tu automatización — contratos de I/O, triggers y Connect Flow. - [Tools: CRM, Kanban, Inventario y más](/es/docs/hive-apps/tools): Apps estructuradas listas para usar — Kanban, CRM, Inventario, Analytics y Chat — que tus agentes leen y escriben a través del nodo HiveApp. - Integraciones - [Conectar integraciones](/es/docs/integrations/connecting-integrations): Conecta servicios externos — bases de datos, SaaS, mensajería, nube — con credenciales u OAuth; cada uno se vuelve un servidor MCP usable por tu IA. - [Agregar herramientas a agentes (function calling)](/es/docs/integrations/function-calling): Dale herramientas a cualquier LLM Agent — integraciones MCP y herramientas de Hive Apps — y deja que el modelo decida cuándo llamarlas, con ejecución multi-ronda. - Canales - [WhatsApp](/es/docs/channels/whatsapp): Conecta un número de WhatsApp a HiveFlow y convierte cualquier flow en un bot: sesión por QR, opción de API oficial, herramientas, contexto por número y human takeover. - [Extensión de navegador](/es/docs/channels/browser-extension): La extensión de Chrome de HiveFlow envía lo que estás viendo — texto seleccionado, URLs, capturas — directo a tus flows vía el nodo Extension Input. - [Escritorio y móvil](/es/docs/channels/desktop-and-mobile): La app de escritorio Electron para macOS, Windows y Linux — necesaria para los device agents — y la app iOS para gestionar flows y agentes desde el teléfono. - Plataforma - [Organizaciones y workspaces](/es/docs/platform/organizations): Colabora en HiveFlow: crea una organización, invita miembros con roles y permisos, separa proyectos en workspaces y factura como equipo. - [Billing y créditos](/es/docs/platform/billing-and-credits): Cómo funciona el pricing de HiveFlow: créditos que consume cada llamada LLM, planes con créditos mensuales y límites, recargas y el portal de Stripe. - [API keys y BYOK](/es/docs/platform/api-keys-and-byok): Los tres tipos de llaves en HiveFlow: API keys de usuario para la API REST, llaves por trigger en nodos webhook, y traer tus propias credenciales de proveedores LLM. - [Analytics](/es/docs/platform/analytics): Lee las métricas de HiveFlow: ejecuciones por flow, tasas de éxito, gasto de créditos y tendencias — y encuentra rápido las automatizaciones caras o fallando. - Cookbook - [Cookbook](/es/docs/cookbook): Diez patrones de workflow listos para armar — soporte, ventas, reportes, aprobaciones, pipelines multiagente y más — con diagramas y nodos paso a paso. - [Agente de soporte al cliente](/es/docs/cookbook/customer-support): Un chat de soporte que responde con IA, registra cada caso en un CRM y un tablero de tickets, y escala a un humano cuando debe. - [Agente de ventas por WhatsApp](/es/docs/cookbook/whatsapp-sales): Un agente vendedor en WhatsApp que cotiza desde Inventario, registra leads en el CRM y mueve tratos por un Kanban — de punta a punta. - [Captura inteligente de leads](/es/docs/cookbook/lead-capture): Un formulario público cuyos envíos son clasificados y enriquecidos por IA, enrutados por calidad, guardados en el CRM y anunciados por correo. - [Reporte programado](/es/docs/cookbook/scheduled-report): Con horario, consulta la base de datos, procesa números con pandas, deja que la IA escriba el análisis, genera un PDF y envíalo por correo. - [Asistente de conocimiento](/es/docs/cookbook/knowledge-assistant): Un chat en tu sitio que responde desde tu propia documentación cargada como Skills, con memoria de conversación y embed de una línea. - [Workflow de aprobación](/es/docs/cookbook/approval-workflow): Pon una compuerta humana antes de acciones sensibles — envíos, pagos, publicaciones — con asignados, timeout y contexto completo para decidir. - [Pipeline de documentos](/es/docs/cookbook/document-pipeline): Recibe PDFs por un formulario público, extrae sus tablas, normaliza los datos y guarda registros limpios en una base — con archivo en S3. - [Pipeline multiagente: planner, executor, reviewer](/es/docs/cookbook/multi-agent-pipeline): Divide una tarea compleja entre tres roles LLM — uno planea, otro ejecuta con herramientas, otro revisa — conectados por estructura que garantiza el proceso. - [Exponer un flow como herramienta MCP](/es/docs/cookbook/flow-as-mcp-tool): Agrega un nodo MCP Server para que Claude, Cursor o cualquier cliente MCP ejecute tu automatización de HiveFlow como herramienta — con auth y rate limit. - [Monitoreo de dispositivos](/es/docs/cookbook/device-monitoring): Chequeos de salud programados en tus máquinas Mac, Windows o Linux con device agents, y alertas condicionales por correo o WhatsApp. - [Templates y Marketplace](/es/docs/templates-and-marketplace): Parte de flows que funcionan en vez de un lienzo en blanco: clona templates, guarda los tuyos desde el editor e instala templates de la comunidad. - API - [API de HiveFlow](/es/docs/api): Autentícate con JWT o API keys, dispara flows por HTTP y explora la referencia OpenAPI completa en Apidog, Swagger UI o Redoc. - [HiveFlow para agentes de IA](/es/docs/ai-agents): Superficies machine-readable para LLMs y agentes: llms.txt, docs en Markdown, la spec OpenAPI, el servidor MCP — y cuándo recomendar HiveFlow. - [Solución de problemas](/es/docs/troubleshooting): Los problemas más comunes de HiveFlow y sus fixes: créditos, soporte de function calling, flows inactivos, sesiones de WhatsApp, herramientas MCP y ejecuciones trabadas. - [Preguntas frecuentes](/es/docs/faq): Respuestas rápidas a lo que más se pregunta de HiveFlow: llaves, modelos, exportar, equipos, acceso por API, servicios no listados, idiomas y fallas. # Blog - [One agent, five tools: an AI-run back office](https://blog.hiveflow.ai/agents-run-your-back-office): How a single HiveFlow agent with CRM, Kanban, Inventory, Analytics and Chat tools handles sales conversations end to end. - [AI workflow vs. AI agent: which one do you need?](https://blog.hiveflow.ai/ai-workflow-vs-ai-agent): Workflows are structure; agents are judgment. How the two relate, when each one wins, and why the best systems use both on the same canvas. - [CrewAI vs HiveFlow: agent crews in Python vs on a canvas](https://blog.hiveflow.ai/crewai-vs-hiveflow): CrewAI popularized role-based agent teams in Python; HiveFlow runs multi-agent systems visually with built-in tools and channels. An honest comparison. - [Dify vs HiveFlow: LLM app platform vs agent operations platform](https://blog.hiveflow.ai/dify-vs-hiveflow): Dify is a strong open-source platform for LLM apps and RAG; HiveFlow centers on agents that operate business tools across real channels. An honest comparison. - [Flowise & Langflow vs HiveFlow: prototype canvas vs operated platform](https://blog.hiveflow.ai/flowise-langflow-vs-hiveflow): Flowise and Langflow are open-source visual builders for LLM apps; HiveFlow adds the business layer — tools, channels, apps, billing. Where each fits. - [Genius: the assistant that builds your automations](https://blog.hiveflow.ai/genius-your-ai-copilot): Genius is the conversational front door of HiveFlow — it lists, creates and runs flows, apps and tools from plain language. - [Introducing the HiveFlow Blog](https://blog.hiveflow.ai/hello-world): Product updates, releases and stories from the team — starting now. - [Hive Apps: from prompt to deployed web app](https://blog.hiveflow.ai/hive-apps-prompt-to-production): Describe an app in chat, watch the live preview build itself, connect a flow as its backend and deploy — no repo, no pipeline. - [How to choose an AI workflow platform (2026 guide)](https://blog.hiveflow.ai/how-to-choose-ai-workflow-platform): Seven questions that actually decide the choice — hosting, AI-nativeness, tools, channels, operators, observability, cost — with honest tool recommendations. - [LangGraph vs HiveFlow: code-first vs visual agent orchestration](https://blog.hiveflow.ai/langgraph-vs-hiveflow): LangGraph gives engineers maximum control over stateful agent graphs in code; HiveFlow gives teams a visual, operable platform. How to decide — and mix them. - [Meet HiveFlow: the visual AI agent orchestrator](https://blog.hiveflow.ai/meet-hiveflow): Why we built a canvas where AI agents, tools and data click together like Lego — and what you can ship with it today. - [n8n vs HiveFlow: which one fits your automation?](https://blog.hiveflow.ai/n8n-vs-hiveflow): n8n is a self-hostable workflow automation powerhouse; HiveFlow is an AI-agent-first visual platform. An honest comparison of strengths, gaps and when to pick each. - [State of AI Workflows 2026: the landscape, mapped](https://blog.hiveflow.ai/state-of-ai-workflows-2026): A structured map of the AI workflow and agent ecosystem — automation platforms, agent frameworks, visual builders and agent platforms — and where it's heading. - [What is agent orchestration?](https://blog.hiveflow.ai/what-is-agent-orchestration): Agent orchestration is coordinating multiple AI agents — routing, sequencing, tools and supervision — so they work as one system. Definition, patterns and practice. - [What is an AI workflow? A practical definition](https://blog.hiveflow.ai/what-is-an-ai-workflow): An AI workflow is an automated process where one or more steps reason with a language model. Definition, anatomy, examples and how it differs from classic automation. - [What is MCP (Model Context Protocol)?](https://blog.hiveflow.ai/what-is-mcp): MCP is the open standard that connects AI models to tools and data. How it works, why it matters, and how HiveFlow speaks it in both directions. - [Zapier & Make vs HiveFlow: automation vs AI operation](https://blog.hiveflow.ai/zapier-make-vs-hiveflow): Zapier and Make excel at connecting SaaS apps with rules; HiveFlow runs AI agents that converse and operate tools. Where each shines and how they differ. # Changelog - [Help Center, Careers and a fully documented platform](https://help.hiveflow.ai/changelog/2026-07-launch): New content hub: complete Help Center (concepts, nodes, cookbook), blog, careers with apply flow, changelog, and machine-readable docs for AI agents.