Skip to Workflow Atlas
ETL PipelineAI systemIntermediate

AI-Powered Image Captioning Workflow

AI-Powered Image Captioning Workflow is a 11-node automation blueprint connecting Google Gemini Chat Model, AI Output Parser, HTTP Request. Inspect its trigger, execution graph, branches, and implementation requirements.

Interactive system map

Follow the runtime. Inspect what supports it.

Solid horizontal paths show main-channel execution. Violet dependencies sit beneath the exact step they configure or empower.

11

Nodes

12

Links

1

Run groups

Connected tools

3 integrations

  • Google AI
  • AI Output Parser
  • HTTP Request

Canvas controls

Fit wide systems, then zoom into any run group.

AI-Powered Image Captioning Workflow execution lanes1 run groups containing 11 workflow nodes and 12 visible links. Solid horizontal lines are main-channel runtime links. Violet dashed lines connect typed dependencies beneath the runtime step they support. Focus any node to inspect its immediate path.When clicking ‘Test workflow’. trigger node, manualTrigger. 0 incoming and 1 outgoing connections.STARTWhen clicking ‘Test workf…TRIGGERMANUALTRIGGER0 IN / 1 OUTGet Image. integration node, httpRequest. 1 incoming and 2 outgoing connections.STEP 02Get ImageINTEGRATIONHTTPREQUEST1 IN / 2 OUTGet Info. integration node, editImage. 1 incoming and 1 outgoing connections.STEP 03Get InfoINTEGRATIONEDITIMAGE1 IN / 1 OUTResize For AI. integration node, editImage. 1 incoming and 1 outgoing connections.STEP 03Resize For AIINTEGRATIONEDITIMAGE1 IN / 1 OUTImage Captioning Agent. ai node, chainLlm. 3 incoming and 1 outgoing connections.STEP 04Image Captioning AgentAICHAINLLM3 IN / 1 OUTMerge Image & Caption. decision node, merge. 2 incoming and 2 outgoing connections.STEP 05Merge Image & CaptionDECISIONMERGE2 IN / 2 OUTCalculate Positioning. transform node, code. 1 incoming and 1 outgoing connections.STEP 06Calculate PositioningTRANSFORMCODE1 IN / 1 OUTMerge Caption & Positions. decision node, merge. 2 incoming and 1 outgoing connections.STEP 07Merge Caption & PositionsDECISIONMERGE2 IN / 1 OUTApply Caption to Image. integration node, editImage. 1 incoming and 0 outgoing connections.OUTCOMEApply Caption to ImageINTEGRATIONEDITIMAGE1 IN / 0 OUTGoogle Gemini Chat Model. Google AI node, lmChatGoogleGemini. 0 incoming and 1 outgoing connections.DEPENDENCY · DEPTH 1Google Gemini Chat ModelGOOGLE AILMCHATGOOGLEGEMINI0 IN / 1 OUTStructured Output Parser. ai node, outputParserStructured. 0 incoming and 1 outgoing connections.DEPENDENCY · DEPTH 1Structured Output ParserAIOUTPUTPARSERSTRUCTURED0 IN / 1 OUT

Signal inspector

Focus, hover, or select a step to isolate its runtime and support links.

Main-channel runtimeTyped dependencytriggeraidecisiontransformintegration
Scroll horizontally to follow every connected flow.11 connected nodesEnter or click to lock · Escape to clear

Runtime breakdown

Read the execution spine before its dependencies.

Each run group separates main-channel steps from the models, tools, retrievers, and data services that support them. Start and outcome labels apply only to runtime nodes.

01

Run Group

When clicking ‘Test workflow’

10 Execution Links2 Support Dependencies
  1. Start

    When clicking ‘Test workflow’

    Trigger

  2. Step 2

    Get Image

    Integration

  3. Step 3

    Get Info

    Integration

    Resize For AI

    Integration

  4. Step 4

    Image Captioning Agent

    AI reasoning

  5. Step 5

    Merge Image & Caption

    Decision

  6. Step 6

    Calculate Positioning

    Transform

  7. Step 7

    Merge Caption & Positions

    Decision

  8. Outcome

    Apply Caption to Image

    Integration

Attached dependencies

These nodes support the runtime above; they are not additional starts or outcomes.

2 nodes
  • Google Gemini Chat Model

    AI reasoning dependency

  • Structured Output Parser

    AI reasoning dependency

Turn the map into a plan

Adapt this blueprint to your company.

Bring the system shape into a workspace, then define the production controls, data boundaries, owners, and delivery sequence around your context.