Harness Generative AI in Your Processes with the Prompt Builder AI Skill

Over the past year, interest in artificial intelligence has surged due to the proliferation of generative AI and large language models. These tools captured imaginations, demonstrating a technology brimming with possibility. While many focused on the potential of these tools, some companies made AI practical. For example, last year, Appian released packaged AI tools for processing content at scale and quickly building interface forms.

Our new generative AI prompt builder AI skill enables you to inject generative AI into your operational processes. Tell the model what to do in plain language, then add the new AI skill to your workflow to drive operational efficiency.

The prompt builder allows developers to design and test prompts to ensure consistent quality. Then, they can add them to a wider process so employees reap the benefits of AI without having to prompt an AI model themselves—and all within the safeguards of corporate policies and Appian’s private AI architecture. 

Let’s walk through an example.

Let’s start with a practical example: extracting personally identifiable information (PII) from a document. Traditionally, this would be an arduous, manual process. But developers can utilize Appian’s new generative AI skill to create a prompt to teach the AI model how to extract PII data, then they can connect it to another automation tool to pull the data out automatically. 

To start, select this option from the AI Skill menu and add some information, such as the prompt’s name.

Next, you can structure the prompt as needed. When writing an effective prompt, be specific. Tell the AI skill exactly what you want it to do and the outcome you expect. For example, consider this prompt: 

"I would like you to determine all PII (personally identifiable information) from a message body and output the result in a JSON format, including keys for the types and their values."

We told the AI skill what to do (extract PII), how to format the output (JSON), and how to map the data (to type and value). This specificity ensures we receive data in the right format to be used elsewhere in our applications and processes.

You can boost precision by providing example inputs and outputs. This teaches the model what to deliver when it receives a request.

Before adding your AI prompt to a wider process, you can test it to ensure you get the expected response. This further ensures accuracy and precision when it runs as part of your automated processes.

Once you’ve built the AI skill, add it into a workflow using the process modeler.

This sends information through a private, generative AI model. After extracting PII, Appian lets you use this newly formatted and structured data in a workflow to update systems or reroute the process.

If 2023 was the year of AI promises, 2024 must be the year of tangible AI benefits. Appian’s new generative AI prompt builder gives developers and designers the power to implement generative AI across the organization. Many employees have used AI and LLMs in their day-to-day; it’s a capability that guarantees more consistent, high-quality results in Appian’s secure private AI architecture. 

Plus, Appian brings AI to life by leveraging strong data and process automation capabilities. This enables you to see real value from AI. These three components—AI, data, and process—make artificial intelligence beneficial at enterprise scales. Instead of using AI in a piecemeal fashion, you leverage it within larger, end-to-end processes consistently, saving you significant time and improving quality.

We have tons of exciting new capabilities, including a preview of our new Case Management Studio, the ability to save and share data fabric analytic reports, and new plug-ins for Appian Portals, to name a few. See demos and more by watching the product announcement webinar.