Unlock the Power of AI in Your Workflows: Introducing OpenAI Pipelines Channel

At Quickbase, we’ve always been committed to helping you solve complex business problems by connecting data, automating workflows, and creating tailored applications—all without writing code.

We’re excited to announce the next step in this journey: the addition of the OpenAI channel in Quickbase Pipelines.

With the AI Smart Builder, we are focused on helping builders create apps and pipelines. However, this integration marks the first milestone in bringing the transformative capabilities of large language models (LLMs) directly into your Quickbase solutions to help you get more out of your data!

This is more than just a new feature—it’s an invitation to experiment, innovate, and reimagine what’s possible within Quickbase.

What Is the OpenAI Channel?

OpenAI GPTs are a powerful tool for generating natural language responses based on prompts. By integrating this functionality via Quickbase Pipelines, you can create workflows that harness the power of AI to transform text, automate repetitive tasks, and interact with unstructured data in ways that were previously out of reach.

How Does It Work in Pipelines?

Pipelines enable seamless integration between Quickbase and third-party tools or systems, allowing you to build robust workflows without coding. With the OpenAI channel, you can now connect to your own OpenAI account and include an AI-powered step in it.

Here are a few examples of how you might use it:

• Content Creation: Use AI to draft emails or internal memos directly from prompts or templates stored in your Quickbase app.
• Summarization: Input lengthy, unstructured text from one or more fields (e.g., meeting notes, incident reports or comments) into the OpenAI endpoint to generate concise summaries or comparisons.
• Fill a form/Make a list: Generate data to populate multiple fields or generate child records based on textual input.

The OpenAI channel connects to the Chat Completion API endpoint of OpenAI.

Why Is This Important?

AI-powered tools like OpenAI’s models are revolutionizing the way we approach problem-solving and process optimization. By integrating LLM capabilities into Quickbase, we’re empowering you to:

• Save Time by automating repetitive tasks and reducing manual effort.
• Enhance Innovation by unlocking new possibilities for streamlining workflows and building smarter solutions.

Get Started: Experiment and Prototype with AI

This integration is just the beginning. We encourage you to start experimenting and prototyping with the OpenAI Chat Completion step in your pipelines. Here are a few tips to get you started:

1. Connect to your OpenAI account in Pipelines.
2. Read the help article and check out the examples below.
3. Identify repetitive or text-heavy workflows where AI could add value, such as document generation or decision support.
4. Test different ways of phrasing user and system prompts to achieve optimal results for your specific needs.

The Future of AI in Quickbase

The OpenAI channel is another step toward broader adoption of AI within Quickbase. We’re actively exploring more ways to bring LLMs and other AI technologies into the platform to empower your business with cutting-edge tools.

As we continue to develop AI-powered features, your feedback and creativity will shape the path forward. We can’t wait to see the innovative ways you’ll use this integration to enhance your workflows and achieve more with Quickbase.

Examples of using the OpenAI channel in Quickbase

Below are a couple of examples that we’ve compiled to help you get started with using the OpenAI channel in your Quickbase workflows.

Example 1 – Create an executive summary for a bid

The first one is from our help article but with a little twist. In the example we show you how you can generate a weekly summary email for each open bid in your Bid Management application (the Bid Management app is available as a Pro App in the Quickbase Exchange).

The pipeline considers all currently open bids in Step A. Then by using OpenAI, it generates a summary email for each open bid and sends it away.

The information used is from the Bid record only.

However, you can choose to upgrade the pipeline a step further yourself.

You can capture all the related child records of a parent record with some Jinja.

In the case of the bid, you can take the Bid Actions/Notes, by adding a “Search Records” step and filtering the records related to the Bid.

Step “a” is the query step to get open bids. Step “b” is the query step to get the related child records (Bid Actions/Notes).

The Jinja below is what we’ve added in the User Message for the OpenAI step.

List of action items and notes in the json: {
    "records": [
        {% for record in b %}
        {
            "action_item_name": "{{ record.Note }}",
            "action_item_status": "{{ record.Status }}",
            }
        }{% if not loop.last %},{% endif %}
        {% endfor %}
    ]
}

This allows you to send the list of child records to OpenAI in the same step as well, giving the LLM more needed information to create a better executive summary.

Example 2 – Fill a form

The second example is a use case that we in Quickbase encounter on a daily basis – how do we analyze and prioritize all of the amazing feedback that we get.

Let’s imagine that we get a lot of emails with feedback and feature ideas. We want to use the power of AI to analyze and categorize each email automatically.

Our pipeline consists of 3 steps:

1. Trigger every time a new email comes in QB (in the form of a record)
2. Analyze the email with OpenAI
3. Update the QB record with the needed information.

Here, we need a way that the information returned from OpenAI is structured in a way that we can utilize it in a pipeline and record it back in our app.

This is where “Structured Output” comes in. We can build how we’d like to see the response from OpenAI.

Here is the User Message as usual.

However, this time I want to get a Structured Output. This means that I need to select and construct it in the pipeline.

I can select Response Format – JSON Schema and build the desired output.

{
       "type":"object",
       "properties":{
          "feature_idea_title":{
             "type":"string"
          },
          "product_team_priority_reasoning":{
             "type":"string"
          },
          "email_content":{
             "type":"object",
             "properties":{
                "customer_severity":{
                   "type":"string",
                   "enum":[
                      "Insignificant",
                      "Minor",
                      "Moderate",
                      "Major",
                      "Catastrophic"
                   ]
                },
                "customer_frustration":{
                   "type":"string",
                   "enum":[
                      "calm",
                      "upset",
                      "furious"
                   ]
                },
                "customer_urgency":{
                   "type":"string",
                   "enum":[
                      "low",
                      "elevated",
                      "severe"
                   ]
                }
             }
          }
       }
    }

This means that I can now refer each one of these objects in the pipeline. OpenAI will transform the entire response to the User Message back to a JSON that the LLM can give values to each set parameter (“feature_idea_title”, “customer_urgency”, etc.)

Read more about OpenAI’s Structured Outputs.

 

Let us know how you’re using AI in your Quickbase solutions—your success stories inspire us to keep pushing the boundaries of what’s possible! Reach out to me at gpeev@quickbase.com

 

* Please note that, as always, the use of AI may provide unexpected results. In addition, we recommend consulting your organization’s internal guidelines prior to sending data to OpenAI. For more on OpenAI’s data privacy policies, visit here. For information on Quickbase’s approach to AI, visit here.