Creating Tasks via Dashboard | Rightbrain Docs | Transform AI Ideas into Production Tasks

Learn how to create, configure, and deploy Tasks using the Rightbrain dashboard. This guide walks you through the visual interface designed for non-technical users to build AI-powered tasks without writing code.

No coding required. The Rightbrain dashboard provides an intuitive interface for creating Tasks with AI assistance at every step.

Overview of the Task Creation Process

Creating a task involves three quick steps:

1. Prompts

Define what your task should do using natural language instructions. Use clear, direct language so the AI understands the task’s goal and context.

2. Inputs / Outputs

Configure input variables and output formatting for your Task. This ensures your data flows in and out of the model consistently and predictably.

3. Config

Name your Task, add a short description, and set model parameters such as temperature and fallback models.

Step 1: Define Your Prompts

Navigate to Create Task and start in the Prompts tab.

User Prompt

The user prompt is where you tell the AI what to do with the input data.

Start with a simple, focused task description

Start with a basic instruction like: "Please conduct sentiment analysis on the {customer_review}"
Variables are automatically detected when you use {brackets}

The Rightbrain Assistant can enhance your prompt automatically:

Click when the assistant appears
Review the improved version
Choose to “Insert Prompt” or customise further

System Prompt (Auto-generated)

The Rightbrain Assistant automatically creates an appropriate system prompt that:

Defines the AI’s role and expertise
Sets additional context for the task
Ensures consistent behaviour

The system prompt is generated behind the scenes. You can view and edit it by expanding the “System Prompt” section if needed.

Step 2: Configure Input/Output (I/O Tab)

Next up is the I/O tab to define your task’s data structure.

Input Configuration

Your input variables are automatically detected from the prompt:

Input Variables

Variables from your prompt appear here automatically

For {customer_review}:

Name: customer_review
Type: String (default)
Click the pencil icon to edit variable settings

Input Format Options

Click the pencil icon next to any input variable to configure its format:

📝 String (default)

Text provided directly when running the task
Use for: reviews, descriptions, questions

🌐 Web Crawl

Automatically fetches content from a web page URL
Use for: website analysis, content extraction

📄 PDFs, PPT and CSVs

Extracts content from uploaded file(s)
Use for: document analysis, invoice processing

🔎 Perplexity Search

Performs a live web search and retrieves a set number of top results
Use for: research tasks, knowledge retrieval, context expansion

To use URL, document or Perplexity search processing, edit your input variable and select the appropriate format option. The system will automatically handle content extraction when the task runs.

Image Handling

Images work differently from other inputs - they don’t require prompt variables:

To accept images in your task:

Don’t add a variable like {image} to your prompt
Instead, navigate to the Images section in the Input tab
Toggle “Required images” if images are required for the Task
Optionally enable “Auto convert images” for optimisation

Example prompt for image tasks:

Analyse this product image and provide quality assessment

Note: No {image} variable needed - the image is automatically included

Output Configuration

Define what information the task should return:

Important: Choose your output modality FIRST before adding any properties. This determines whether your task generates text, images, or audio.

📝 Text Output
- Default structured data format
- Returns formatted text with defined properties
- Use for: analysis, classification, extraction, and summarisation
🖼️ Image Output
- AI-generated visuals created from prompts
- Produces image files directly in your results
- Use for: visualisations, designs, artwork, product photos
🎧 Audio Output
- AI-generated audio or speech files
- Converts model output into spoken responses or sounds
- Use for: voice assistants, accessibility, sound generation

For Text Outputs - Define Properties

When using Text output mode, you can define structured data:

Manual Setup

AI Generation

Click “Add Property” to manually define outputs:

Enter a property name (e.g., “sentiment”)
Add a description
Select the type (String, Number, Boolean, etc.)
Indicate whether the property is an Array
For options, add the allowed values using “Add Enum”

Schema vs Visual Views:

Schema: Shows the technical structure
Visual: Displays a user-friendly preview

For Image and Audio outputs: No additional properties are needed. The task will generate the appropriate file type based on your prompts and inputs.

Output Types Guide

For Text output mode, choose the appropriate type for each field:

Output Types Guide

For Text output mode, choose the appropriate type for each field:

Type	Description	Examples
String/Text	Text fields for freeform responses	Summaries; classifications; extracted text
Enum	String with fixed choice options	positive/negative/neutral; low/medium/high; approved/rejected
Boolean	True/False values for binary decisions	is_valid; requires_review; matches_criteria
Number	Decimal values for numeric data (supports floats)	Scores (0–100.0); percentages; prices
Integer	Whole numbers only	counts; quantities; rank positions

Step 3: Configure Settings

Click the Settings tab to finalise your task configuration.

Task Information

Task Name

Required

A descriptive name for your task

Auto-generated example: “Sentiment Analyser”

Requirements: Letters, numbers, spaces, and hyphens only

Description

Explain what your task does and when to use it

Auto-generated example: “Analyses customer reviews to determine overall sentiment, providing classification and justification”

Model Selection

Primary Model

Required

Select from available models based on your needs:

Vision-capable models for image analysis tasks
Fast models for simple classification
Advanced models for complex reasoning

Fallback Model

Choose a fallback model in case your primary model fails. If a request to your primary model fails 4 times, tasks will be routed to your back up model.

Temperature Setting

Temperature

Defaults to 30%

Controls response consistency (0-100%):

0-30%: Highly consistent (classifications, extractions)
30-50%: Balanced (analysis, summaries)
50-100%: Creative (content generation, brainstorming)

AI Pre-configuration: The Rightbrain Assistant automatically generates a name and description, and sets optimal values for model and temperature based on your task type. You can always update these during task creation or later.

Using the Rightbrain Assistant

The assistant appears at key moments to help optimise your task:

Prompt Enhancement

When you write a basic prompt, the assistant can:

Add structured sections (Goal, Context, Input Parameters, Processing Steps)
Include best practices for prompt engineering
Clarify instructions and expected outputs

Example Enhancement:

From:

Please conduct sentiment analysis on the {customer_review}

To:

Goal:
Conduct a sentiment analysis on a provided customer review 
to determine the overall sentiment expressed.
Context:
Sentiment analysis involves evaluating written text to identify 
the emotional tone, such as positive, negative, or neutral.
Input Parameters:
{customer_review} - The text of the customer review to be analyzed
Processing Steps:
1. Read and understand the customer review
2. Identify emotional indicators and tone
3. Classify the overall sentiment
4. Provide reasoning for the classification

Output Generation

Click “Generate Output” and the assistant will:

Analyse your prompt to understand the task goal
Suggest appropriate output fields with types
Add helpful descriptions for each field

Once configured, click “Create Task” to start testing your task instantly.

Next Steps

Run Your Task

Test and deploy your task

API Integration

Connect via API when ready