With the dramatic rise of generative AI platforms like ChatGPT, Scribe, and DALL-E2, prompt engineering has been receiving more and more attention. When it comes to generative AI, “garbage in” often equals “garbage out.” The opposite is also true; well-constructed and tailored prompts lead to more usable and contextual results.
Prompt engineering also has interesting applications for UX. When users access an application, they often run into questions, concerns, or uncertainty about what to do next. And while a chatbot-driven support window can be helpful for handling these issues, it requires users to know what to ask and to actually submit their query.
With prompt engineering, however, Nintex Apps builders, formerly Skuid, can tailor our platform’s AI capabilities to produce a more intuitive UX that guides users programmatically — no active user input necessary. In this, we’ll share some of the UX advantages of prompt engineering over open-ended chatbots, as well as several potential use cases. We’ll also go through examples of how to actually leverage prompt engineering in Nintex.
Benefits of prompt engineering over open-ended chatbots
Today’s software users have high expectations when it comes to UX. They want consumer-grade app experiences from business applications and employee portals, complete with easy navigation, personalization, auto-fill, and real-time feedback. Unfortunately, chat boxes can’t deliver on all of these fronts.
Prompt engineering, however, offers enhanced capabilities around UX. Let’s go through four of its key advantages in more detail.
1. Personalized user interactions
Unlike chatbots that wait for user queries, AI-driven prompts proactively guide users based on previous actions. For example, a sales rep might access their dashboard and receive suggestions for next steps based on prior interactions, like following up with a client about a specific product. Using the Nintex platform, builders can implement parameters like “recently viewed products” or “last interaction date” to construct prompts like “Follow up with Client X about Product Y.”
2. Real-time feedback and guidance
While chatbots react to user inputs, prompt engineering provides immediate feedback as a user completes a task, preventing potentially costly errors. Let’s say a rep fills out a Nintex form with the required information. Behind the scenes, parameters such as “input length” or “data type” can be checked to construct prompts like “Ensure the email format is correct.”
3. Dynamic content presentation
AI-driven insights adapt the content presentation to different display preferences and prior interactions. For example, a Nintex Apps portal might rearrange content based on a user’s most-accessed sections or most-viewed pages. These parameters can then prompt users to take a specific action, like “Check out our latest article on Topic Z.”
4. Enhanced decision-making
Chatbots produce data, but AI-driven prompts offer actionable suggestions that can streamline decision-making processes. Consider a financial dashboard built in Nintex Apps that suggests investment strategies. Market data parameters, such as “rising stocks” or “volatile sectors,” can create user prompts like “Consider investing in Sector A due to recent growth.”
Prompt engineering in Nintex
Let’s go through two practical examples of how Nintex Apps builders can leverage prompt engineering to enhance user experience.
Nintex forms with real-time feedback and guidance
Consider a Nintex Apps form where users input key customer data, including email addresses, phone numbers, and purchase history. As users enter this information, Nintex’s prompt engineering provides feedback.
To construct the prompt, builders will need to set the appropriate Nintex Model properties, action framework, and JavaScript:
- Nintex Model Properties: The form’s underlying model can have properties that define acceptable data formats. An example would be a regex pattern for email validation.
- Action Framework: As users input data, Nintex’s action framework can trigger specific responses based on conditions. If a user’s input doesn’t match the email regex pattern, for instance, an action can be triggered to display a prompt that helps them fix the error. For example, if the input is “john.doe@domain,” but the user hasn’t added “.com” yet, Nintex’s action framework might trigger a prompt saying, “Did you mean john.doe@domain.com?” Or, if a user inputs a phone number without the country code, a prompt might suggest, “Consider adding the country code for clarity.”
- JavaScript (if needed): For more complex validations or dynamic prompts, builders can integrate custom JS snippets. For example, if a user inputs a purchase date in the future, JS can be used to validate the date and trigger a prompt suggesting a correction, such as, “The purchase date seems to be in the future. Please check and correct.”
Menu actions within a Nintex application
Let’s go through one more example of prompt engineering in action. Nintex Apps builders can leverage our OpenAI integration to create a list of suggested user actions displayed as a right-click menu. These actions are pre-built prompts to help end users make productive requests to the OpenAI’s gpt-3.5-turbo-0301 model and can be grouped in categories like Generate, Transform, or Analyze.
In this Nintex Apps Labs experiment, we created a menu of user actions using a Task object from Salesforce, including “Generate Offer Letter,” “Generate Email Response,” “Increase Urgency,” and “Personalize Message.” However, you can configure the action sequence to be connected to any object or data source you have connected in Nintex Apps — the possibilities are endless!
Use AI to your app’s advantage
No matter what type of application you’re building with Nintex, our platform’s AI capabilities and OpenAI integration can help you transform your UX for the better. Here are some resources to get you started:
