Swiftask
  • Quick start
  • Key concepts
    • AI Tools Hub
    • Agents
    • Knowledge base
    • Skills
    • Projects
    • Automation
  • AI tools hub
    • Introduction
    • Chat interface
    • Tokens
    • List of AI features
    • AI suggestions
    • FAQ
  • Agents
    • Introduction
    • Create an agent step by step
    • How to evaluate your agent
    • Multi-agents
    • Widget
    • Share agent
    • FAQ
  • Knowledge base
    • Introduction
    • Data connectors
      • Rich text
      • PDF File
      • Azure Document Loader
      • YouTube
      • Apify Dataset
      • PowerPoint File
      • Excel File
      • DOCX File
      • SQL Database
      • REST API
      • JSON File
      • CSV File
      • SQL Database Query
      • Website
      • Webpage
      • Sitemap
      • Dropbox files
      • Google drive files
    • Create a knowledge base
    • Attach Knowledge base to your agent
    • Share knowledge base
    • FAQ
  • Skills
    • Introduction
    • Skills library
      • Webpage Content Parsing
      • GitLab File Creation
      • Browsing with Perplexity
      • Open API
      • Retriever data from external sources
      • GitHub pull request diff retriever
      • GitHub pull request comment
      • Export table to Excel
      • Export text to PDF
      • GitHub file content
      • GitHub pull request info
      • OpenDataSoft
      • Agent as Skill
      • Swiftask AI recommandation
      • LinkedIn Share
      • Prismic migration create
      • Github create file
    • Create a new skill
    • Attach skill to your agent
    • FAQ
  • Projects
    • Introduction
    • Create a project
    • Generate task
    • Task AI chat
    • Project agent
    • FAQ
  • Automation
    • Introduction
    • Create an automation
  • Workspace admin
    • Introduction
    • Invite collaborators to join your workspace
    • Subscription renewal and Credit explanation
    • Purchase credits
    • Referral
    • Subscription plan & token distribution
    • Cancel subscription /Manage payment method
    • Personnal data security
    • SSO For enterprise
  • Use cases & Tutorials
    • Chat with multi-AI
    • Chat with PDF file
    • Import data - Webpage
    • How to generate an image on Swiftask
    • Import data (Azure Document Loader) - PDF
    • How to generate videos on Swiftask
    • Transform your ideas into videos with LUMA AI
    • Upgrade subscription plan
    • How to create an agent? step by step
    • Create AI agents for your business
    • Integrate external API in your agent
    • Create a professional landing page in 5 minutes
    • How to automate your blog content creation with an AI agent
    • How to evaluate your AI agent
    • How to create a Community Manager agent
  • Developer
    • List of AI and agents accessible via API
    • Access AI and agent through API
    • OpenAI SDK
  • Support & Social network
  • Changelog
    • May 29, 2025 - Enhanced Voice Input & AI Transcription
    • May 25, 2025 - Enhanced AI Models & Improved User Experience
    • May 25, 2025 - Mistral Medium 3 & Small 3 Integration
    • May 19, 2025 - Audio Recording Interface Improvements
    • May 19, 2025 - Third-Party Authentication Management
    • May 17, 2025- Enhanced Google Drive Synchronization & Privacy Improvements
Powered by GitBook
On this page
  1. Agents

How to evaluate your agent

PreviousCreate an agent step by stepNextMulti-agents

Last updated 19 days ago

Here is how to evaluate agents in Swiftask:

1

Access the "Agents" section.

2

Select your agent, "Edit agent"

3

Once in the Agent setting interface, navigate to the "Evaluation" section.

4

Click on the "Add" button.

5

You have two options:

  • Choose "Add an evaluation" to manually enter the questions and expected answers.

  • Or select "Import an evaluation from a CSV file" to load data from a file.

6

For the "Add an evaluation" option:

fill out the form with the question and the expected answer from the agent. You can also click on the "Add to evaluation" button, which appears at the bottom of the chat with the agent when you test it. This will automatically generate a form with your question as the prompt and the agent's response as the expected answer.

7

Once you have saved the questions and answers, check those on which you want to evaluate your agent, then click on "Run Evaluation".

You will then receive a score based on:

✅ Relevance of responses ✅ Factual accuracy ✅ Context precision

This feature will allow you to ensure high-quality responses and improve user experience.