We have a library of over 500 professionally crafted prompts for many use cases. The aim is to save your time by offering keyword search for the relevant prompt instead of you spending time drafting it yourself. In many cases it is best practice to use prompts in combination with Agents. Here is how to access the prompt library:
You can search for prompts by entering keywords into the search box or ou can filter them by tags. Then just click "Use now" in the Agent card. The prompt will then be automatically loaded into the chat input box on the main page.
You can add your own prompts and agentic workflows by clicking the blue "Add prompt" button at the top right of the Prompt Library.
In the context of conversational AI and models like NIONIUM AI, a "prompt" is the initial message or input that a user provides to the model to generate a response or output. A prompt usually serves three purposes in guiding the model's output:
- Task: The specific action or type of content you're asking the model to generate. For example, "write a poem" or "explain a concept."
- Instructions: Any specific instructions or constraints that accompany the task to ensure that the output meets the user's expectations. This might include tone, format, or specific points to include.
- Role: A persona or character that the model adopts while responding, which can help define the voice, perspective, and mannerisms of the reply. This can be a professional role, like a lawyer, doctor, or teacher, and helps to tailor the response according to the expected knowledge and behavior of that role.
In the NIONIUM AI framework, we have delineated the concepts of Task and Instructions from that of Role, recognizing that a single role or character may carry out a multitude of specialized tasks. Conversely, specialized tasks are typically not interchangeable among various characters.
Consider the following illustration:
A lawyer, conceptualized here as a professionally trained AI character, is capable of crafting a diverse array of legal agreements. To accommodate this multidisciplinary function, we have compiled a comprehensive library of characters. Additionally, to address the variety of agreements and other specialized tasks associated with distinct characters, we have systematically categorized an array of engineered prompts in a separate "Prompt Library." This structured organization streamlines the workflow and ensures precise, role-appropriate task execution.
As an example, for generating a loan agreement, a properly engineered prompt to a model adopting the role of a lawyer could look like this:
"Generate a detailed loan agreement suitable for a private unsecured loan between individuals, ensuring that it includes clauses related to the loan amount, repayment schedule, interest rate, default conditions, and governing law provisions. Please draft this in a formal style appropriate for legal documents and in accordance with common legal practice."
By combining the task (generate a loan agreement) with the instructions (detailing what clauses to include) and the role (a lawyer), the model is better equipped to create a loan agreement that is realistic, professionally structured, and highly usable.
Using professionally engineered prompts is essential for the following reasons:
- Accuracy and Relevance: Well-crafted prompts help to achieve more accurate and relevant outputs by providing clear instructions and context. This reduces ambiguity and guides the language model to respond within a specific framework.
- Efficiency: Professionally engineered prompts can reduce the need for multiple iterations of questions and answers, saving time and resources.
- Quality Control: By creating prompts that are specific, detailed, and aligned with the role or character the model adopts, the output can be managed to a higher quality standard, ensuring that it is appropriate for its intended use.
- User Experience: An effective prompt enhances user experience by yielding responses that align with users’ expectations, thereby increasing their satisfaction with the tool.
When a prompt is used in combination with an AI Agent, such as a lawyer, the benefits are even more pronounced:
- Contextual Authority: Assuming the role of a lawyer, the language model might use more formal language and reference legal principles or the format typical of legal documents.
- Task Orientation: If the task is to create a specific contract, like a loan agreement, the AI, leveraging the lawyer persona, would focus on the legal requirements and clauses typically included in such agreements.
- Targeted Output: The combination of a precise prompt with a relevant persona ensures that the model's output is not just generalized advice but is specific and actionable, like drafting a clause or explaining a legal concept.
As an example, for generating a loan agreement, a properly engineered prompt to a model adopting the role of a lawyer could look like this:
"Generate a detailed loan agreement suitable for a private unsecured loan between individuals, ensuring that it includes clauses related to the loan amount, repayment schedule, interest rate, default conditions, and governing law provisions. Please draft this in a formal style appropriate for legal documents and in accordance with common legal practice."
By combining the task (generate a loan agreement) with the instructions (detailing what clauses to include) and the role (a lawyer), the model is better equipped to create a loan agreement that is realistic, professionally structured, and highly usable.