Complete Guide to Working with AI Assistants in AYETO

Welcome to the guide that will walk you through all the features for working with AI assistants in the AYETO platform. You will learn how to create, configure, and effectively use your own AI assistants.

Contents

1. What is an AI Assistant?
2. Accessing Assistants
3. Creating a New Assistant
4. Template Gallery
5. Editing an Existing Assistant
6. Knowledge Base (RAG)
7. Task Assignment
8. AI Tools
9. Sharing an Assistant
10. Export and Import
11. Working with Conversations
12. Advanced Settings
13. Tips and Tricks

What is an AI Assistant?

An AI assistant in AYETO is your personalized AI helper that combines:

• A selected language model (e.g. GPT-4, Claude, Gemini)
• Custom instructions defining personality and behavior
• A knowledge base with your documents (optional)
• AI tools for extended functionality (web search, image generation, etc.)

Thanks to assistants, you can have specialized AI for different tasks—from programming to customer support to creative writing.

Accessing Assistants

Opening the assistants dialog

1. Click the Assistants icon in the main menu
2. A dialog opens with a list of all your assistants
3. You can search using the search field

Assistant overview

Each assistant card displays:

• Name and avatar
• Chat model (language model for conversations)
• Image model (model for image generation)
• Assistant ID (for advanced use)
• A sharing icon (if the assistant is shared)

Creating a New Assistant

Creating a new assistant takes place in a wizard with several steps:

Step 1: Basic Information

💡 Tip: The description is required if you plan to use the assistant in an AI team!

Step 2: Model Selection

• Chat model – the main language model (we recommend models with tool support)
• Image model – model for image generation

Step 3: Instructions

Here you define the assistant’s personality and behavior:

You are a friendly technical assistant specializing in Python programming.
You respond concisely and factually. Always provide code examples.

⚠️ Warning: Instructions must not be empty and have a length limit.

Step 4: AI Tools

Select the tools the assistant will have available:

Basic tools:

• Web search
• Image generation
• File handling
• And more...

Advanced tools:

• Code execution
• Integration with external services
• Special functions

Step 5: RAG Settings

RAG (Retrieval-Augmented Generation) allows the assistant to work with a knowledge base from your documents:

• AI tools – enable Knowledge Base Search

Advanced RAG settings:

• Chunk size – size of document segments (0 = default)
• Top K – number of relevant segments to retrieve (0 = default)

Step 6: Uploading Documents

Supported formats:

• PDF, DOCX, TXT
• Markdown files
• And other text formats

Step 7: Review and Create

Review the summary of settings and click Create Assistant.

Template Gallery

Don’t want to start from scratch? Use ready-made templates!

How to use a template

1. In the assistants dialog, click Gallery
2. Browse templates by category or tags
3. Select a template and click Use Template
4. The assistant will be created with preconfigured values

Filtering templates

• All – complete list
• Recommended – selected templates
• By tags – filtering by categories

Editing an Existing Assistant

Accessing settings

1. In the assistants list, click the menu (three dots)
2. Select Edit
3. Or click the settings icon in the conversation

Settings tabs

General

• Name and description
• Model selection
• Instructions for the assistant

Tools

• Overview of all available AI tools
• Grouped by categories
• Option to enable/disable individual tools

Knowledge Base

• The knowledge base is used via the tool 'Knowledge Base Search'
• Management of uploaded documents
• Adding YouTube transcripts

Conversations

• Overview of conversations with the assistant

Tasks

• Overview of tasks the assistant is supposed to perform

Advanced

• Model temperature – creativity of responses (0 = conservative, 1 = creative)
• Assistant memory – long-term notes for the assistant

Knowledge Base (RAG)

What is RAG?

RAG (Retrieval-Augmented Generation) combines:

1. Searching relevant parts from your documents
2. Generating an answer based on the found information

Uploading documents

1. Go to the Knowledge Base tab
2. Click Upload New Documents
3. Select files (maximum size and quantity are limited)
4. Wait for processing

YouTube transcripts

You can also add transcripts from YouTube videos:

1. Go to the Add YouTube Transcript tab
2. Paste the video URL
3. Click Add

Document management

• View all uploaded documents
• Delete individual documents
• Bulk delete all documents

Task Assignment

What are tasks for?

Tasks are used to automate assistant activities.

1. For a task, we specify the task prompt
2. We define the method of triggering the task
3. The task execution results are stored in the form of a conversation for the task
4. This conversation begins with the task prompt and continues with the assistant’s response

Task definition

1. Task Name - the name by which we will refer to the task
2. Description - a brief description of the task
3. Prompt - instructions explaining to the agent what it should do within the task

Configuring the task trigger

1. Manual - the task is launched only manually - by pressing the Run Now button.

2. Scheduled - the task launches automatically at the scheduled time - select the time zone - enter the date and time

3. Recurring - the task launches repeatedly according to the specified schedule - the same settings format is used as for the cron service - in the Schedule selection, examples of settings are available - the Custom item opens editing for a custom cron expression - select the time zone

Advanced task settings

1. Status - active - the task can be started - paused - the task does not run
2. Maximum number of runs - a limit on how many times the task can run

Notifications

1. Notify on completion - after the task is completed, a notification about its result is sent
2. Notify on failure - a notification is sent in case of failure

Delete button

• deletes the task and its history

History button

• displays the execution history

Run Now button

• starts the task
• also starts scheduled tasks, i.e. outside the schedule

Task execution history

• after a task runs, information about its result is stored in Execution History
• individual runs are available in the list
• after clicking, the result of the run is available, i.e. the conversation created by executing the task
• the Full Conversation button displays the entire conversation in the conversation context
• this conversation contains the task prompt and the assistant’s response, i.e. the task execution
• you can continue the conversation further there
• ⚠️ however, the conversation is stored only in the task history, i.e. it is not visible in the normal list of conversations

AI Tools

Overview of tools

AI tools extend the assistant’s capabilities. They are divided into groups:

Tool requirements

⚠️ To use tools, the selected model must support function calling (tools capability).

Sharing an Assistant

Sharing with other users

1. In the assistant settings, go to the Sharing section
2. Click Share with User
3. Enter the user’s email
4. Save changes

Sharing permissions

Sharing allows other users to:

• Use the assistant in conversations
• Copy the assistant (creates their own copy)
• Export the assistant

Export and Import

Exporting an assistant

1. In the assistant menu, select Export
2. Choose options: - ✅ Include knowledge base - ✅ Include memory
3. Click Download Export File

Export creates a JSON file with all settings.

Importing an assistant

1. In the assistants dialog, click Import
2. Select the JSON file
3. Enter a name for the imported assistant
4. Choose options: - ✅ Import knowledge base - ✅ Import memory
5. Click Import

Copying an assistant

For quickly creating a copy:

1. In the assistant menu, select Copy
2. Enter a new name
3. Choose whether to copy the knowledge base
4. Confirm

Working with Conversations

Starting a conversation

1. In the assistants list, click Chat
2. Or create a new conversation and select the assistant

Chat interface

• Header – information about the assistant and model
• Message history – previous conversations
• Input field – writing messages, uploading files

Features in the conversation

• Image generation – switch to image generation mode
• Attachments – upload files and images
• Voice input – dictate messages
• Export conversation – save as PDF or DOCX

Advanced Settings

Model temperature

Affects the creativity of responses:

Assistant memory

Long-term memory for storing:

• User preferences
• Important information
• Context across conversations

Memory is editable in advanced settings.

Strict RAG mode

When enabled:

• The assistant responds ONLY based on documents
• It does not add information from the model’s general knowledge
• Ideal for precise, verifiable answers

Tips and Tricks

📝 Effective instructions

✅ Good:
"You are a Python expert. Respond concisely. Always provide code examples."

❌ Bad:
"Be a good assistant."

🎯 Specialization

Create multiple assistants for different tasks:

• Programmer – technical help
• Copywriter – creative texts
• Analyst – working with data

📚 RAG optimization

• Upload relevant, high-quality documents
• Split large documents into smaller parts
• Regularly update the knowledge base

🔧 Model selection

• For complex tasks: GPT-4, Claude 3 Opus
• For quick responses: GPT-3.5, Claude 3 Haiku
• For coding: Models with tool support

💾 Backup

Regularly export important assistants in case of:

• Accidental deletion
• Transfer to another account
• Sharing with colleagues

Need help?

• 📧 Contact support
• 📖 Visit the FAQ section