Welcome to Luna, your organisation’s personalised AI assistant! At Yopla, our mission is to make business better by fostering more open, thriving, and sustainable organisations. Luna is here to help you increase free time, reduce stress, and support your team’s growth.
By leveraging Luna's capabilities, you can streamline tasks, improve productivity, and gain insights tailored to your specific needs.
Getting the most from Luna: prompt engineering
To maximise the effectiveness of your Luna assistant, crafting your prompts carefully is key. Here’s how you can do it:
1. Be clear and specific
Luna performs best when given clear, specific instructions. Avoid vague or broad prompts to get the most accurate and useful responses.
Example 1: Instead of asking, “Help me with spreadsheets,” ask, “How can I create a pivot table in Excel to analyse sales data from the last quarter?”
Example 2: Instead of saying, “Tell me about programming,” specify, “Can you explain how to use Python for web scraping?”
2. Provide context and background information
Giving context helps Luna understand the scope and relevance of your query, leading to more precise and tailored responses.
Example: Instead of asking, “How do I configure this tool?” provide context: “I am setting up Jira for our project management. How do I configure workflows for a software development project?”
3. Break down complex queries
Complex questions can be overwhelming. Breaking them down into smaller, manageable parts ensures Luna can address each aspect effectively.
Example: Instead of asking, “How do I set up a server?” break it down: “What are the steps to install Ubuntu on a server? How do I configure Apache for hosting a website?”
4. Use keywords and phrases
Including relevant keywords and phrases helps Luna identify the main focus of your query, enhancing the accuracy of the response.
Example: Instead of asking, “What’s new in tech?” specify, “What are the latest features in the latest version of Docker?”
5. Expect qualifying questions
Luna may ask qualifying questions to ensure it provides the best advice. Being prepared for these can help streamline the interaction.
Example: If Luna asks for clarification, provide detailed responses to help it give the most accurate advice: “What specific version of the software are you using?”
6. Experiment and iterate
Prompt engineering is an iterative process. Experimenting with different phrasings and structures helps you discover what works best for your needs.
Example: If a prompt doesn’t yield the desired response, try rephrasing it or adding more detail. For instance, “How do I secure a database?” might be rephrased to, “What are the best practices for securing a MySQL database against SQL injection attacks?”
Conclusion
Effective prompt engineering enhances your interaction with Luna, allowing you to leverage its full capabilities. By being clear, specific, and iterative, you can ensure Luna provides you with the most accurate and useful information tailored to your needs.
For more information and advanced tips keep in touch with team@yopla.co.uk. Let’s harness the power of Luna together and take your organisation to the next level!
FAQ's
Why does Luna sometimes make mistakes?
Luna is designed to provide highly accurate information, but like all AI, it’s not perfect. Luna operates at a degree-level accuracy, meaning it gets things right about 82% of the time or higher. However, because Luna is trained on a wide range of data sources, including public data and detailed knowledge bases, conflicts or variations in that data can sometimes lead to unexpected answers. This is why Luna might not always give the best answer on the first try, and why it’s important to refine your prompts or ask follow-up questions to get the most accurate results.How should I think of Luna?The quality of Luna’s responses improves with the detail and context you provide in your prompts. While Luna can enhance your work by providing insights, analysis and streamlining tasks, it’s not a substitute for experience and human knowledge. The best results come from combining Luna’s AI capabilities with your expertise, guiding your assistant with clear and detailed instructions to get the most accurate and useful answers.
How accurate is Luna?
Luna is very accurate, with a performance comparable to degree-level expertise—meaning it gets things right about 82% of the time or higher. However, no AI can guarantee 100% accuracy. Measuring accuracy is really challenging as many of the questions we ask of AI are subjective or rely on interpretation. The key to getting the best results is to provide clear, specific prompts and be prepared to refine your questions if needed.
What is the difference between a prompt and a question?
A prompt is a carefully crafted instruction that guides Luna to produce the best possible response. It often includes context, keywords, and specific details to narrow down the information Luna provides. A question is typically more straightforward and may be broader, leading to a less focused answer. Using prompts effectively can help Luna give you the most accurate and useful information.
What sources of knowledge does Luna use?
Luna is trained on a combination of detailed knowledge sources and public data, making it a versatile tool for a wide range of queries. These sources include technical manuals, best practice guides, knowledge bases, proprietary libraries and other authoritative resources relevant to the tasks at hand. The breadth of these sources allows Luna to provide informed answers, but it also means that conflicting information can sometimes result in unexpected responses.
What about conflicts in data?
Like a person, Luna interprets the information it has access to. This interpretation can sometimes lead to conflicts, hallucinations, or mistakes, especially when the data sources contain contradictory information. Luna might provide an answer that reflects one interpretation over another, or it might generate unexpected responses if the data is unclear or conflicting. This is a natural limitation of AI, and it's why refining your prompts and providing clear context is essential to getting the best results.
How can I improve the accuracy of Luna’s responses?
To get the most accurate results from Luna, try the following tips:
Be clear and specific in your prompts. The more detail you provide, the better Luna can tailor its response.
Provide context for your queries. This helps Luna understand the relevance and scope of your request.
Break down complex queries into smaller, manageable parts.
Use relevant keywords and phrases to guide Luna’s focus.
Experiment and iterate by refining your prompts if the initial response isn’t what you expected.
Can Luna give different answers to the same question?
Yes, Luna might provide different answers to the same question depending on the prompt, the context provided, and the specific data sources it draws from. This variation is a natural part of how AI models work, especially when dealing with complex or ambiguous queries. To get the most reliable answer, it’s a good idea to refine your prompts and provide additional context.
Is there anything I can do to achieve 100% accuracy with Luna?
While it’s challenging to achieve 100% accuracy with any AI, you can get closer by crafting precise prompts, providing as much context as possible, and being open to refining your queries based on the responses you receive. Remember, Luna is a tool to assist you, and while it can significantly enhance your productivity, it’s always good to double-check critical information.