1. Understanding the Basics
Callout: This section is geared toward those who are new to AI. If you want to skip to the meat and potatoes of how the new features work, bump down to “2. Application Scenarios: When and How to Use AI in Tines”
What is AI?
AI or Artificial intelligence is the simulation of intelligence delivered by computers. These computers have been fed a ton of data to learn and begin to understand. After a certain point, you can ask the computer a question, and it uses what it’s learned to begin to provide a better, more understood answer.
If we consider one of the most primitive computers, in this case, a calculator, we can input numbers and ask it to provide a solution—something as simple as 1+1 or more complex like the square root of 153.
As the computers for companies like OpenAI, Anthropic, and Bard get more and more data to learn from, they begin to develop a higher level of comprehension when asked a question.
An example could be giving it a news article, asking it to provide a summary, looking at logs, and extrapolating any data that seem to have abnormalities or stand out. Without getting incredibly deep, this technology allows us a new generation of data processing and computer comprehension.
What is a Prompt?
A prompt is the way we ask questions to any given AI. It is the set of instructions in our request.
Example 1:
Intellect and wisdom are two qualities that often intersect but have distinct meanings and applications.
Intellect refers primarily to the mental capacity and ability to reason, analyze, and acquire knowledge. It involves cognitive abilities such as the capacity to learn new information, apply logic, solve problems, make judgments, and comprehend complex ideas. Intellect is often associated with formal education, scholarly pursuits, and the accumulation of factual knowledge. It's about the sharpness of the mind and the efficiency with which it processes and integrates information.
Wisdom, on the other hand, involves a deeper level of understanding and the ability to apply knowledge and experiences judiciously to make sound decisions. Wisdom is characterized by insight, good judgment, and the foresight to predict outcomes or the empathy to understand people deeply. Unlike intellect, which can be seen as raw computing power, wisdom is more about how you use that power in real-life situations. It includes the ability to see the bigger picture and often involves ethical considerations, compassion, and a sense of balance.
In essence, if intellect is knowing a tomato is a fruit, wisdom is not putting it in a fruit salad. Wisdom often requires not just intellectual skills but emotional maturity and life experience, allowing individuals to navigate complex life and social challenges more effectively. While intellect can be developed through education and learning, wisdom is typically honed through experience and reflective practice.
If we want a briefer and simpler response, we can provide that in our prompt!
Example 2:
Intellect is the ability to acquire and process knowledge, while wisdom is the discerning application of that knowledge in practical life.
The way we ask this question is typically referred to as Prompt Engineering.
What is Prompt Engineering?
As we develop our request to an AI platform, we want to be mindful of how we ask. It’s the same as saying you were hungry to a friend versus telling them you want to get pizza together because you’re hungry. More context and information can provide us with a more refined answer.
A few key elements are worth noting: the tone, the role, and the style.
Tone:
In your prompt, you can guide it to sound friendly and informative or even in the voice of Shakespeare. This indicates the words developed in the response. Think of how you’d get an email from a coworker vs a friend.
Role:
Developing a role for the prompt can give a bit of a “starter pack” for guidance. Something like:
You are a security engineer with 10 years of experience. You currently work on teaching new members of your organization the basics. When responding, break things down into easy-to-understand concepts.
If you apply something like this at the beginning of your prompt, you give the context of someone who may have that type of background. Remember, AI is built on a ton of information, so it can be pretty wise!
Style:
Style can be a very concise response or a long, detailed summary that provides insight. The style can also inform what the response may come as.
2. Application Scenarios
Let's examine some ways to leverage AI's power as we develop our stories. Below are a few general categories of workflows where you can implement AI.
When and How to Use AI in Tines
Uplifting a Manual Process
Previously, you might have needed to manually analyze data to understand changes or discrepancies, similar to tasks involved in data governance.
Dynamically Creating Objects
Ad-hoc creation of Slack blocks.
Generating Slack bots with customizable features, such as two buttons and a header.
Developing HTML tailored to specific use cases.
Triage Information
Ingesting data and evaluating it against predefined conditions.
Applying specific formatting when conditions are met.
Reformatting Data
Cleaning HTML dynamically.
Converting data formats, including CSV and Markdown, normalizing strings to JSON, HTML to JSON, etc.
Summarizing Data
Processing and summarizing critical data, identifying high-risk elements with AI-driven insights.
Simplify Data Processing
Utilizing automatic transform mode to reduce the number of necessary actions in a workflow.
Tips for working with AI
Simplifying Instructions: Start with straightforward tasks (e.g., converting data formats, summarizing information).
Breaking Down Tasks: Approach complex processes incrementally to manage impacts effectively.
Example:
Only show items that are high-priority
Capitalize names
Only show the status, email, id, and name fields
Setting Strict Rules: Specify exact transformations for consistent elements (e.g., identifiers).
Convert to JSON; don't show any other extra text
Utilizing Keywords and Phrases: Employ conditional and directive words to refine AI responses (e.g., "If", "Always", "Provide the response in JSON").
Example
If there are no items that show as malicious, summarize the data. If there are malicious items, provide as much context as possible about each item and recommendations from the perspective of a security expert.
Always provide a summary before showing the data in a markdown table.
Changing the Order: Adjust the sequence of requirements in your prompt to avoid confusion. For instance, specify the need for JSON formatting at the end rather than the beginning.
Hands-On With Tines Tools
1. Event Transform Action - Automatic Mode
Let's take a look at how to use the automatic mode
Highlights
Setup and Usage
Insert and configure the event transform action.
Iterate with the guide to refine outputs.
When to Use: Ideal for consistently formatted data needing logical transformations.
2. AI Action
Let's take a look at how to use the AI action.
Highlights
Setup and Usage
Deploy AI action within a story.
Craft and refine prompts based on the dynamic nature of the data.
When to Use: Best for dynamic data requiring adaptable, intelligent processing.
Custom AI providers
If you'd like to use a different AI provider, you can go into the settings to select different options. This provides a way to use AI outside of Tines. Below are a few quick screenshots to showcase how to access the settings.
Conclusion and Iteration
We are excited to see what you can whip up with these new features to our workflow platform. You'll find more resources in our Tines Explained with additional examples. It is also worth noting that when you use the AI action, there will be easy-to-access templates to help you start/build some inspiration!
Feedback and Continuous Improvement
Invite feedback on each module through interactive elements (e.g., thumbs up/down).
Adapt training content based on user responses and emerging needs.