Step 2: Route Message Types and Map Text

Route each incoming message type through the correct parsing path and map text for the AI agent.

1. In Detect Message Format, verify the four rules evaluate {{ $json.body.data.messageType }} against conversation, imageMessage, audioMessage, and documentMessage.
2. Confirm the fallback output is named Message not supported and connects to Unsupported Message Reply.
3. In Map Text Fields, set text to {{ $json.body.data.message.conversation }} so conversation messages go directly to the agent.

Execution Flow: Incoming Webhook Trigger → Detect Message Format → Map Text Fields → Orchestration Agent.

Step 3: Configure Media Retrieval and Conversion

Set up base64 retrieval and binary conversion for audio, images, and documents so they can be processed by Gemini.

1. Configure Fetch Audio Base64, Retrieve Image Base64, and Fetch Document Base64 with Operation set to get-media-base64 and Message ID to {{ $json.body.data.key.id }}.
2. In Convert Audio Binary, set Operation to toBinary, Source Property to data.base64, and Binary Property Name to data_1.
3. In Convert Image Binary and Convert Document Binary, set Operation to toBinary and Source Property to data.base64.

Execution Flow: Detect Message Format → Retrieve Image Base64 → Convert Image Binary → Image Analysis and Detect Message Format → Fetch Audio Base64 → Convert Audio Binary → Audio Transcription.

Execution Flow: Detect Message Format → Fetch Document Base64 → Convert Document Binary → Document Summary.

Credential Required: Connect your Evolution API credentials for all Evolution API nodes (5 nodes handle inbound and outbound media/messages).

Step 4: Set Up Gemini Processing and Mapping

Configure Gemini to transcribe audio, analyze images, and summarize documents, then map the resulting text for the agent.

1. In Audio Transcription, set Resource to audio, Input Type to binary, and Binary Property Name to data_1.
2. In Image Analysis, keep the long analysis prompt in Text and set Resource to image with Input Type binary.
3. In Document Summary, set Text to What is this document about? Please provide a summary of the subject matter. Be as brief as possible. and Resource to document.
4. Map AI outputs in Audio Text Mapping and Document Text Mapping with text set to {{ $json.content.parts[0].text }}.
5. In Image Text Mapping, set Imagem_description to {{ $json.content.parts[0].text }} and image_caption to {{ $('Detect Message Format').item.json.body.data.message.imageMessage.caption }}.

Credential Required: Connect your Google Gemini credentials in Audio Transcription, Image Analysis, and Document Summary.

Step 5: Configure the AI Orchestration and Memory

Set up the agent that decides how to respond and uses memory and tools for calendar actions.

1. In Gemini Chat Model, keep default settings to serve as the language model for Orchestration Agent.
2. In Short-Term Memory, set Session Key to {{ $('Incoming Webhook Trigger').item.json.body.data.key.remoteJid }}, Session ID Type to customKey, and Context Window Length to 8.
3. In Orchestration Agent, set Text to {{ $json.text }} and keep the defined system message instructions.
4. Attach Calendar MCP Tool as an AI tool in Orchestration Agent with Endpoint URL https://hexagom-n8n.cloudfy.cloud/mcp/calendar.

Credential Required: Connect your Google Gemini credentials in Gemini Chat Model.

Short-Term Memory and Calendar MCP Tool are AI sub-nodes; they are configured through Orchestration Agent, so ensure the parent node is properly set up.

Step 6: Configure Calendar MCP Server and Google Calendar Tools

Expose calendar tools through the MCP server so the agent can create, retrieve, modify, and delete events.

1. In MCP Calendar Server, set Path to calendar.
2. In Create Calendar Event, set Start to {{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Start', ``, 'string') }} and End to {{ /*n8n-auto-generated-fromAI-override*/ $fromAI('End', ``, 'string') }}.
3. In Create Calendar Event, set Calendar to {{ '[YOUR_EMAIL]' }}, and map Summary and Description using the provided $fromAI expressions.
4. In Retrieve Calendar Events, set Operation to getAll, Time Min to {{ /*n8n-auto-generated-fromAI-override*/ $fromAI('After', ``, 'string') }}, and Time Max to {{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Before', ``, 'string') }}.
5. In Remove Calendar Event and Modify Calendar Event, set Event ID to {{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Event_ID', ``, 'string') }} and set Calendar to {{ '[YOUR_EMAIL]' }}.

Credential Required: Connect your Google Calendar credentials for Create Calendar Event, Retrieve Calendar Events, Remove Calendar Event, and Modify Calendar Event.

Step 7: Configure Response Dispatch

Send the agent’s response back to the original sender.

1. In Dispatch Response, set Remote Jid to {{ $('Incoming Webhook Trigger').item.json.body.data.key.remoteJid }}.
2. Set Message Text to {{ $json.output }}.

Credential Required: Connect your Evolution API credentials in Dispatch Response and Unsupported Message Reply.

Step 8: Test and Activate Your Workflow

Validate end-to-end behavior and activate for production.

1. Use Execute Workflow and send test POST requests to Incoming Webhook Trigger for conversation, image, audio, and document messages.
2. Confirm successful execution: text routes through Map Text Fields, media routes through the conversion and Gemini nodes, and responses are delivered by Dispatch Response.
3. Verify calendar actions through Create Calendar Event, Retrieve Calendar Events, Modify Calendar Event, and Remove Calendar Event when the agent decides tools are required.
4. When ready, toggle the workflow to Active so the webhook runs in production.