menu_bookScale Vibe Agent / Documentation
Overview

Welcome to Scale Vibe Agent

Scale Vibe Agent is a visual AI automation system that lets you build and deploy intelligent agents with workflows connecting AI models to practical tools — without requiring coding expertise.

What can you build with Scale Vibe Agent?

  • sync_altAgents that execute tasks across multiple platforms simultaneously
  • psychologyContext-aware decision-making flows with branching logic
  • upload_fileKnowledge-grounded assistants using your own uploaded documents
  • scheduleScheduled automations triggered by time, events, or webhooks
  • hubMulti-agent networks with intelligent routing between specialists
codeDevelopersInfrastructure-free workflow building
business_centerBusiness TeamsAutomate repetitive tasks
support_agentSupport TeamsKnowledge-based AI agents
campaignSales & MarketingPersonalised lead qualification

How It Works

Scale Vibe Agent follows a four-step process — from visual design to live deployment — all within a single canvas-based interface.

dashboard_customize
STEP 01

Visual Flow Builder

Design your workflow on a canvas-based node editor. Drag, drop, and connect triggers, agents, tools, and logic nodes without writing a single line of code.

smart_toy
STEP 02

Configure AI Agents

Select your AI model (GPT-4, Claude, Gemini), write a system prompt to define behaviour, and attach tools or a knowledge base to supercharge capabilities.

cable
STEP 03

Connect Tools

Integrate with Gmail, Slack, Google Sheets, databases, web search, and 100+ other services via pre-built connectors using OAuth, API keys, or bot tokens.

monitoring
STEP 04

Deploy & Monitor

Launch your agent and track performance in real time — analytics, execution logs, token usage, and cost tracking all in one dashboard.

Quickstart Guide

Create your first AI-powered workflow in under 10 minutes with these seven steps.

1

Sign Up & Log In

Register via email/password or Google Sign-In at vibe.globalagentstudio.ai. Existing users simply log in with credentials.

2

Explore the Dashboard

The dashboard is your control hub — it shows analytics, quick-action buttons, and navigation to agents, flows, templates, and credentials.

3

Create Your First Flow

A flow is a visual workflow connecting agents, triggers, and tools. Choose a template (e.g. "Customer Support Bot") or start from scratch by dragging nodes onto the canvas.

4

Configure Your Agent

The Agent Node is the intelligence engine. Write a system prompt defining behaviour, select an AI model (GPT-4o for reasoning, GPT-4o-mini for speed), and optionally add tools or a knowledge base.

5

Test Your Flow

Click "Test" or "Run", use the chat interface to enter sample inputs, and review the Terminal View for detailed execution logs.

6

Save & Deploy

Save your workflow, choose privacy settings (Private or Public), and optionally configure a trigger — Webhook, Schedule, or Manual.

7

Monitor & Iterate

Return to the dashboard to review analytics, check activity logs, and refine your prompts based on real usage patterns.

tips_and_updatesPro tip

Start with a template from the Template Library — they come pre-wired with agents, triggers, and tools so you can see a working flow in seconds before customising it.

Key Features

Everything you need to build production-grade AI agents out of the box.

hub

Multiple AI Models

Switch between OpenAI, Anthropic, and Google providers on a per-node basis.

extension

100+ Integrations

Connect tools via the MCP protocol — Gmail, Slack, Notion, GitHub, Salesforce and more.

database

RAG Knowledge Base

Upload PDFs, DOCX, TXT, and CSV files to give agents access to your private data.

account_tree

Smart Routing

Rule-based branching with Decision Logic nodes or AI-powered routing via Supervisor nodes.

bolt

Flexible Triggers

Launch agents via chat, webhooks, cron schedules, manual execution, or embedded widgets.

monitoring

Analytics Dashboard

Real-time performance monitoring, execution history, token usage, and cost tracking.

Triggers

The Trigger is the starting point of every workflow. It defines how your agent gets activated. Every flow requires a Start Node that specifies trigger rules.

Five trigger types are available:

chat

Conversational Chat

Most Common

Activates when users send text messages. The system captures input as the user_message variable. Enable "Fresh Start" to treat each message as an independent conversation — useful for FAQ bots.

play_circle

Manual Execution

Dev / Testing

Triggers the flow once with no user input when you click Run. Best for one-off tasks or testing logic during development.

schedule

Scheduled Automation

Cron-based

Runs automatically at configurable intervals — daily, weekly, or custom frequency — at an exact time with timezone support. Ideal for news digests, reports, and periodic monitoring.

webhook

External Webhook

HTTP

Generates a unique, public URL that triggers your agent via HTTP requests. Save the flow first to activate. Supports query parameters for input and includes a URL regeneration option for security.

widgets

Embeddable Widget

Website

A fully customisable chat widget for website integration. Configure theme, accent colour, avatar, header text, suggested questions, auto-open delay, and session tracking via userId.

Agents

The Agent Node is the intelligence engine of your workflow — a Large Language Model configured for specific reasoning tasks. The effectiveness of an agent depends almost entirely on its System Prompt.

terminalSystem Prompt

The foundational instruction that defines your agent's persona, behaviour, and guardrails. Think of it as a high-priority configuration file that constrains what the model does and how it responds.

system_prompt_example.txtcontent_copy
# Define the core identity
You are a Senior DevOps Engineer specialising in
production-ready Python infrastructure.

# Constraints
- Only provide production-grade, tested solutions
- Never expose internal IP addresses or secrets
- Escalate critical errors immediately
- Maintain a professional, technical tone at all times

layersModel Selection

Choose high-reasoning models for complex tasks or faster, cost-efficient options for simpler work.

GPT-4oRecommended
Claude 3.5 SonnetHigh Reasoning
Gemini 1.5 ProMultimodal
GPT-4o-miniLow Latency

thermostatTemperature

Controls the randomness and creativity of outputs.

Deterministic0.0 – 0.3
Data extraction, coding, factual Q&A
Balanced0.4 – 0.6
General purpose workflows
Creative0.7 – 1.0
Brainstorming, content generation

buildTool Augmentation

Connect your agents to external services and APIs. When activated, the agent automatically identifies needed tools, selects appropriate methods, and executes actions.

languageWeb Search
codeCode Exec
apiCustom API

Tools & Integrations

Tools are pre-built connectors that allow AI agents to interact with external services and APIs. Scale Vibe Agent organises them into ten categories.

search

Search & Research

Tavily, Perplexity, SERP

mail

Communication

Gmail, Slack, Discord

task_alt

Productivity

Notion, Jira, Asana

folder

File Management

Google Drive, Dropbox

people

CRM & Sales

Salesforce, HubSpot, Apollo

campaign

Marketing

Twitter, LinkedIn, Mailchimp

database

Data Storage

Postgres, MongoDB, Pinecone

code

Developer

GitHub, Jira, webhooks

share

Social Media

Twitter, Instagram, LinkedIn

handyman

Utilities

Web scraping, forms, parsers

Authentication Methods

verified_userOAuth 2.0Single-click via Google/Microsoft
keyAPI KeysService-provided credentials
tokenBot TokensFor Slack, Discord, etc.
lock_openNo AuthPublic utility tools

Logic & Routing

Scale Vibe Agent provides two distinct node types for controlling workflow branching and decision-making.

rule

Decision Logic Node

Deterministic

Rule-based routing via precise conditions: Variable + Operator + Value. Creates a visual split:

True PathCondition satisfied
False PathCondition fails

Best for black-and-white scenarios: keyword filtering, error handling, user segmentation.

psychology

Supervisor Node

AI-Driven

Instead of matching keywords, the Supervisor analyses the intent behind a request. Configure it with natural language routing instructions:

"Route technical support questions to
the Engineer Agent, and billing inquiries
to the Sales Agent."

Requires a reasoning-capable model (GPT-4, Claude 3.5+). Best for complex multi-agent workflows with unpredictable user intent.

Knowledge Base (RAG)

Retrieval-Augmented Generation transforms your agent from a closed model into an open-book system. Instead of hallucinating answers, the agent searches your uploaded documents for relevant context before responding.

Setup Process

1Drag a Knowledge Base node onto the canvas
2Draw a connection from your Agent Node to the Knowledge Base node to grant access permissions
3Upload data via file upload, URL, or Google Drive integration

Data Input Methods

upload_file
File UploadPDFs, Word docs, text files from your computer
link
URL AdditionWeb links the system scrapes and indexes
drive_folder_upload
Google DriveCloud documents with account connection

Supported Formats

PDFReports, contracts, presentations
DOCXEditable Word documents
CSV / JSONStructured data and catalogs
TXTNotes, logs, plain text
Max file size: 100 MB per file

File Processing States

ProcessingIndexing in progress
ReadyAvailable for queries
FailedUpload or processing error

If a file fails, delete it and re-upload. Common causes: file size exceeds 100 MB or corrupted file.

AI Model Selection

Scale Vibe Agent supports multiple frontier LLMs. Switch providers on a per-agent basis to optimise for reasoning, speed, or cost.

ModelProviderBest ForSpeed
GPT-4oRecommendedOpenAIComplex reasoning, multi-step tasksMedium
GPT-4o-miniOpenAISimple tasks, high-volume flowsFast
Claude 3.5 SonnetAnthropicLong context, nuanced writingMedium
Gemini 1.5 ProGoogleMultimodal, document analysisMedium

General Settings

Configure flow-level settings that apply to the entire workflow — including privacy, error handling, and execution behaviour.

lock

Privacy

Set flows to Private (only you) or Public (shareable via link or embed).

bug_report

Error Handling

Define what happens when a node fails — retry, skip, or halt the flow.

memory

Memory Mode

Enable persistent memory so agents remember context across sessions.

history

Execution History

View past runs with timestamps, inputs, outputs, and token usage.

Creation Wizard

The Creation Wizard guides you through setting up a new agent from scratch, step by step.

1Name your agent and choose an icon for easy identification
2Select the trigger type (Chat, Schedule, Webhook, Widget, or Manual)
3Add and configure one or more Agent nodes with your model and system prompt
4Connect tools and optional Knowledge Base nodes
5Add logic nodes if your flow requires conditional branching
6Test with sample inputs and review the execution trace
7Deploy by saving and activating your chosen trigger

My Agents

The My Agents dashboard gives you a full view of every agent you've built — with quick actions for each.

edit

Edit Flow

Reopen the canvas to modify nodes

play_circle

Run / Test

Execute with a test input

monitoring

View Analytics

Tokens used, run history, costs

content_copy

Duplicate

Clone the flow as a starting point

share

Share

Generate a shareable embed or link

delete

Delete

Permanently remove the agent

Ready to start building?

Launch the platform and build your first AI agent — free, no credit card required.

rocket_launchOpen Scale Vibe AgentTalk to an Engineer
Get Free DemoBuild Free Agent