Avatar Pipeline

High-Level System Overview

Shows the main components: Client onboarding inputs flow through the N8N/Cloud Functions processing layer, interact with AI services, and output weekly video deliveries.

flowchart LR
    subgraph Input["Client Onboarding"]
        CRM["CRM
(GoHighLevel)"]
        Videos["Training
Videos"]
        Form["Intake
Form"]
    end

    subgraph Processing["AI Processing Pipeline"]
        N8N["N8N
Workflows"]
        GCF["Google Cloud
Functions"]
    end

    subgraph AI["AI Services"]
        EL["ElevenLabs
Voice Clone"]
        HG["HeyGen
Avatar Video"]
        OAI["OpenAI
Script Writing"]
    end

    subgraph Output["Weekly Delivery"]
        Drive["Google
Drive"]
        Slack["Slack
QC & Delivery"]
    end

    Input --> Processing
    Processing <--> AI
    Processing --> Output

Client Journey: Onboarding to First Videos

Timeline showing how a new client moves from intake form submission through AI training to weekly video production.

flowchart TB
    subgraph Week1["Week 1: Onboarding"]
        A1["Client submits
intake form"] --> A2["Client uploads
5+ training videos"]
        A2 --> A3["Videos processed
& stored"]
    end

    subgraph Week1b["Week 1: AI Training"]
        B1["ElevenLabs creates
voice clone"] --> B3["Voice training
completes"]
        B4["HeyGen creates
avatar(s)"] --> B5["Multiple 'looks'
available"]
    end

    subgraph Week1c["Week 1: Content"]
        C1["AI generates
60 scripts"] --> C2["Scripts reviewed
& queued"]
    end

    subgraph Week2["Week 2+: Production"]
        D1["5 scripts selected
for the week"] --> D2["Audio generated
from scripts"]
        D2 --> D3["Videos created
with avatar"]
        D3 --> D4["QC review
in Slack"]
        D4 --> D5["Delivered to
client Friday"]
    end

    A3 --> B1
    A3 --> B4
    A1 --> C1
    B3 --> D1
    B5 --> D3

    style Week1 fill:#e1f5fe
    style Week1b fill:#fff3e0
    style Week1c fill:#f3e5f5
    style Week2 fill:#e8f5e9

The Job Queue System

How the Google Sheets-based job queue works: triggers add jobs, the Runner Poller picks them up, executes workflows, and moves jobs through pending → in_progress → completed states.

flowchart LR
    subgraph Trigger["Triggers"]
        Cron["Cloud Scheduler
(every 1-20 min)"]
        Webhook["Webhooks
(CRM, Slack)"]
    end

    subgraph Queue["Google Sheets Job Queue"]
        Pending["Pending
Jobs"]
        InProgress["In Progress
(max 15)"]
        Completed["Completed"]
    end

    subgraph Workers["N8N Workers"]
        Runner["Runner Poller
(1-min)"]
        Workflows["Sub-Workflows"]
    end

    Cron --> |"Check for work"| Pending
    Webhook --> |"Add new job"| Pending
    Pending --> |"Pick up job"| Runner
    Runner --> |"Execute"| Workflows
    Runner --> |"Update status"| InProgress
    Workflows --> |"On complete"| Completed
    Workflows --> |"Enqueue next"| Pending

    style Pending fill:#fff9c4
    style InProgress fill:#bbdefb
    style Completed fill:#c8e6c9

API Integration Map

How the N8N orchestration layer connects to external APIs: ElevenLabs for voice, HeyGen for avatars, OpenAI for scripts, CloudConvert for media processing.

flowchart TB
    subgraph N8N["N8N Orchestration"]
        Orch["Workflow
Engine"]
    end

    subgraph Voice["Voice Pipeline"]
        V1["Upload Audio
Samples"]
        V2["Train Voice
Clone"]
        V3["Generate
Speech"]
    end

    subgraph Avatar["Avatar Pipeline"]
        A1["Upload Training
Videos"]
        A2["Create Avatar
Looks"]
        A3["Generate
Videos"]
    end

    subgraph Content["Content Pipeline"]
        C1["Extract Form
Data"]
        C2["Generate
Scripts"]
    end

    EL["ElevenLabs API"]
    HG["HeyGen API"]
    OAI["OpenAI API"]
    CC["CloudConvert API"]

    Orch --> Voice
    Orch --> Avatar
    Orch --> Content

    V1 --> EL
    V2 --> EL
    V3 --> EL

    A1 --> CC
    A2 --> HG
    A3 --> HG

    C1 --> OAI
    C2 --> OAI

    style EL fill:#ff7043,color:#fff
    style HG fill:#7e57c2,color:#fff
    style OAI fill:#26a69a,color:#fff
    style CC fill:#5c6bc0,color:#fff

Weekly Production Cycle

Timeline showing the typical weekly production flow from Monday queuing through Friday delivery.

flowchart LR
    subgraph Monday["Mon"]
        M1["Queue 5 scripts
per client"]
        M2["Start audio
generation"]
    end

    subgraph TueWed["Tue-Wed"]
        T1["Generate avatar
videos"]
    end

    subgraph Thursday["Thu"]
        TH1["QC review
in Slack"]
        TH2["Prepare delivery
folders"]
    end

    subgraph Friday["Fri"]
        F1["Deliver videos
to clients"]
    end

    M1 --> M2 --> T1 --> TH1 --> TH2 --> F1

    style Monday fill:#e3f2fd
    style TueWed fill:#fff3e0
    style Thursday fill:#f3e5f5
    style Friday fill:#e8f5e9

Component Architecture

System context showing users (Client, Internal Team), the central Avatar Pipeline platform, and external service integrations.

flowchart TB
    subgraph People["Users"]
        client["Client
Business owner receiving
weekly avatar videos"]
        team["Internal Team
QC and operations"]
    end

    subgraph Core["Avatar Platform"]
        pipeline["Avatar Pipeline
Automated video
production system"]
    end

    subgraph External["External Services"]
        crm["GoHighLevel
CRM & onboarding"]
        elevenlabs["ElevenLabs
Voice cloning"]
        heygen["HeyGen
Avatar videos"]
        openai["OpenAI
Script generation"]
    end

    client -->|"Submits intake form"| crm
    crm -->|"Triggers onboarding"| pipeline
    pipeline -->|"Clones voice"| elevenlabs
    pipeline -->|"Creates avatars & videos"| heygen
    pipeline -->|"Generates scripts"| openai
    team -->|"Reviews & approves"| pipeline
    pipeline -->|"Delivers videos weekly"| client

    style client fill:#4CAF50,color:#fff
    style team fill:#2196F3,color:#fff
    style pipeline fill:#9C27B0,color:#fff
    style crm fill:#FF9800,color:#fff
    style elevenlabs fill:#ff7043,color:#fff
    style heygen fill:#7e57c2,color:#fff
    style openai fill:#26a69a,color:#fff

Workflow Dependency Graph

Shows how N8N workflows connect: Phase 1 (Onboarding) feeds into Phase 2 (Weekly Production), which outputs to Phase 3 (Delivery & QC).

flowchart LR
    subgraph Onboarding["Phase 1: Onboarding"]
        direction TB
        Init["InitializeClient"] --> Scripts["GenerateScripts"]
        Init --> Upload["UploadMedia"]
        Upload --> Convert["ConvertToAudio"]
        Upload --> Avatar["CreateAvatar"]
        Convert --> AddVoice["AddVoiceSample"]
        AddVoice --> Train["TrainVoice"]
    end

    subgraph Production["Phase 2: Weekly Production"]
        direction TB
        Weekly["Weekly Schedule"] -.-> Queue["QueueWeeklyVideos"]
        Queue --> S2A["ScriptsToAudio"]
        S2A --> GenVid["GenerateSingleVideo"]
    end

    subgraph Delivery["Phase 3: Delivery & QC"]
        direction TB
        GenVid2["Videos Complete"] --> QC["HandleSlackQC"]
        Prepare["PrepareDelivery
(Thu)"] --> Deliver["DeliveryBySlack
(Fri)"]
    end

    Train --> Queue
    GenVid --> GenVid2

    style Init fill:#e3f2fd
    style Train fill:#fff3e0
    style Queue fill:#e8f5e9
    style Deliver fill:#fce4ec
    style Weekly fill:#fff9c4