Understanding Input Prompts

In this segment of the prompt guide, we'll explore what are input prompts for Freddy Copilot by examining examples of both successful prompts and common pitfalls. You'll gain valuable insights into how to harness the power of Freddy effectively.

Freddy Copilot for Developers works really well with Freshworks context, however, it is a Generative AI powered solution. Hence it falls short in certain areas. We are working towards continously improving on them. Here is a quick list of same.

1. Lack of Common Sense: Freddy does not have a deep understanding of common-sense knowledge. It responses based on patterns in the data it is trained on, and its knowledge is limited to Freshworks context.

2. Incorrect Information: Freshworks App ecosytem has evolved over years and is under constant change and we are working towards providing best in class DevX. Due to this Freddy can sometimes produce information that is incorrect, or outdated, especially in integration use cases. It's important to fact-check information obtained from Freddy when something doesn't work.

3. Difficulty in Controlling Output: While you can guide the Freddy's output with prompts, achieving precise control over the generated content can be challenging especially in complex use cases involving ambiguity. Freddy may occasionally generate unexpected or off-topic responses.

4. Verbose Responses: Freddy may produce lengthy, verbose responses when a concise answer is needed. It can require additional effort on your side to extract the relevant information.

5. Dependency on Input: We are making Freddy smarter and more capable with each release. However, the quality of the output depends on the quality of the input prompt. Crafting an effective prompt can be challenging, and slight changes in wording can produce different results.

   For example:

   Freshworks provides UI library for building Freshworks Apps that are in sync with product look and feel. The use of word Crayons can ensure that the generated output follows Freshworks UI standards always.

   Freshworks recommends to build app with app logic inside app.js for handling frontend actions as you can use items such as Interface Methods, Events method etc. Due to the generative nature of Freddy, it can at times generate results otherwise.

   Further, the use of word additional can clarify whether to replace existing fields or to append additional fields. Check the tabs to understand more.

6. Lack of Real-Time Knowledge: Freddy do not have real-time access to current events or live data. Hence it's responses are based on its training data, which may become outdated

7. Resource Intensive: Setting up higher context settings value for Freddy or using large input prompts require significant computational resources, utilising higher number of tokens per response.

We understand that no Generative AI system is perfect and so are we. However, we are working towards improvement and increasing accuracy of Freddy and there is a long way ahead. We need your support to improve and grow Freddy to match your needs. At end of each prompt output, we provide an upvote and downvote options, we urge you to use them appropriately and provide us your valueable feedback as we continue to refine Freddy.