This guide provides a comprehensive walkthrough for creating and configuring Ejento AI assistants tailored to your specific use case. It covers the key aspects required to define your assistant's purpose, behavior, interaction style, and reasoning capabilities.By following this guide, you will be able to design assistants that are not only intelligent and goal-driven but also aligned with your product or team's requirements.
Overview#
Naming Strategy
Establishing a clear, relevant name that reflects the assistant’s functionality.
Assistant Description
Writing a concise yet impactful description that communicates what the assistant does and who it's designed for.
Defining Roles and Responsibilities
Structuring the assistant’s core functions and expected behaviors to ensure clarity and focus.
Custom Instructions and Behavioral Guidelines
Crafting effective prompts and guidelines to shape how the assistant responds and interacts with users.
Conversation Starters
Adding initial interaction examples to help users engage with the assistant naturally and effectively.
Selecting a Reasoning Mode
Choosing the most appropriate reasoning pattern (e.g., ReAct, Reflection, CodeAct) to match the assistant's task type and decision-making style.
1. Navigate to the assistants page from the sidebar, and Click on Add Assistant#
2. Open the "Select Team" dropdown to select the team for Assistant#
3. Select your team for the assistant here#
4. Create a New Team#
If you want to add your assistant under a new team, click Create new team and follow the step-by-step guide of creating a new team here
5. After selecting the Team, Now select the Project for the assistant here#
6. Within Teams, you can either select an existing project or create a new one.#
7. If you want to create a new project, click on Create new Project and enter your project name.#
8. Give your project a suitable and meaningful name.#
9. Click on "Create Project"#
10. Click on "Add Assistant"#
Your assistant will be created within the selected team and project.
11. Click on this icon if you want to set or update your assistant’s image#
12. You’ll find three options here: Upload Image, Generate Image, and Remove Image.#
Use Upload Image to add a custom image, Generate Image to create one for your assistant, or Remove Image to delete the current image.
13. Click on “Generate Image” then Enter a prompt describing how you want your assistant’s image to look.#
14. Click on “Generate” to create an image based on your entered prompt#
15. If you’re satisfied with the generated image, click “Select”; otherwise, update your prompt and try again#
16. Give Assistant a suitable name#
17. Write a Description about what the Assistant does#
18. Define the Assistant’s Role#
19. Click on the Assistant Model section to choose your preferred model for the assistant#
20. Here you’ll see all available models choose the one that best fits your needs#
21. Click on Add Custom Instruction#
Custom behavior instructions enable fine-tuning to handle specific scenarios or requirements. These instructions dictate how the assistant should act or what it must take care of in its responses.
22. Give Custom Instructions#
23. Click on Save#
24. Select the Visibility of your Assistant#
25. Click on More Options#
26. Conversation starters are disabled by default; enable the toggle if you want to add them#
27. Add Conversation Starters for the Assistant#
Provide a few example questions or prompts that users can ask to begin interacting with the assistant.
28. Select a Reasoning Pattern for Your Assistant#
Choose the most suitable reasoning mode based on how you want your assistant to think and respond.
For a complete guide on available reasoning modes and how to choose the right one, visit:
Reasoning Modes Guide 29. Choose How Your Assistant Thinks#
Here, We selected CodeAct as the reasoning mode because our assistant is designed to generate, explain, and interact with code intelligently.
Attachment: Allows users to upload and share files directly with the assistant.
Web Search: Enables the assistant to access and provide up-to-date information from the internet.
RAG (Retrieval-Augmented Generation): Lets the assistant generate responses using your uploaded documents or connected data sources.
33. To add a new custom tag, simply type it and click “Add”#
34. You can enable caching to help your assistant respond faster#
Caching stores certain previous responses to improve speed, but in some cases, it may slightly affect personalized accuracy.
35. The Public Link toggle allows you to share or embed your assistant publicly#
36. Click on Create assistant#
37. Chat with the assistant to observe the instruction being followed#
38. Here you can see the assistant’s conversation starters#
You can either choose a question from these conversation starters or type your own query to begin chatting.