How to Build AI Assistants in Knack Using OpenAI and Flows

Introduction:

Knack users—from SMB businesses and nonprofits to consultants and startups—rely on no-code tools to build custom apps that manage their data and simplify workflows without needing technical expertise. These apps help replace spreadsheets and manual processes with flexible, user-friendly solutions.

By integrating OpenAI with Knack Flows, you can now embed intelligent AI Assistants directly into your apps. These assistants handle questions, automate tasks, and provide personalized support, making your apps more dynamic and powerful.

This guide will walk you through building AI-powered virtual helpers inside your Knack apps, helping you save time, reduce manual work, and deliver better experiences for your team and customers.

If you are just interested in using ChatGPT with a simpler ChatGPT "chat completion" action and not using an AI assistant, please see this article: https://learn.knack.com/article/openai-chatgpt-connector-tutorial-for-flows.

SMART HIRING MADE EASY: CREATING AI ASSISTANTS IN KNACK FLOWS FOR RECRUITMENT

Recruiting top talent is vital, but often time-consuming and resource-intensive. Manually reviewing every candidate’s application can slow down the process and introduce inconsistency. With Knack Flows’ direct AI Assistant integration powered by OpenAI, you can automate candidate evaluation, getting detailed insights and suitability scores instantly, right within your app. This smart automation not only speeds up your hiring pipeline but also ensures a consistent, objective screening process.

In this guide, we’ll walk through creating an AI-powered recruitment flow in Knack that analyzes applications, scores candidates, and helps your HR team make data-driven decisions with ease.

---

First, let's generate our API keys and create our AI assistants in OpenAI:

Generating your API key in OpenAI:

To connect OpenAI with Knack Flows, you'll need to create a connection using your API keys. These can be generated in your OpenAI account dashboard: https://platform.openai.com/api-keys.

Creating an AI assistant in OpenAI.

Once you have your API key saved, go to the Assistants tab and click + Create:

On the right-hand side of the screen, you will see several fields we can use to instruct and refine your AI Assistant:

AI Assistant Settings: 

• Name – A user-friendly label for your assistant (e.g., “HR Screener”).
• Assistant ID – The unique identifier of this assistant, you will see this next to the Name of the Assistant when configuring flows later.
• System Instructions – Define the assistant’s behavior (e.g., “You are an experienced HR recruitment assistant who evaluates job applications.”).
• Model – Choose an AI model like GPT-4o or GPT-4 Turbo for best results, 3.5 for cost efficiency.
• Tools – Optional features like File Search, Code Interpreter, or Functions.
  • (File Search) could be used here, for example, if you uploaded a document version of the Job Application details and requirements to the file search field, the AI will also evaluate that Job Application Document while analyzing your Flows application data, making a very powerful, precise response.
  • (Code Interpreter) (also called Advanced Data Analysis) allows it to perform calculations, analyze data, or write and run code.
  • (Functions) enable the assistant to call external APIs or custom tools you define, making it possible to trigger dynamic actions beyond text responses
• Temperature / Top P 0.0 to 1.0 – Controls how creative or focused the responses are (lower values = more predictable, higher values = more random).
• Response Format – Choose how the assistant replies (usually “text” or “JSON”).
  
  

This is my *HR Screener AI assistant, along with the system instructions I used to summarize job application data collected through our Knack forms.

Now you have everything you need to build your own AI-powered automation assistant.

You can build multiple assistants within your flows, each handling a specific task. By passing data to them or between them, you effectively create a team of AI assistants — analyzing, refining, and even triggering other actions or workflows.

It’s a powerful way to scale intelligence across your app without writing a single line of code.

In the next section, we will show this process by passing this AI HR Screener response to another AI assistant we created called "Application Rank Assistant" to rank the application on a scale of 1 - 10.

We can then update our Knack application record with the detailed analysis and the applicant rank, which can help your HR team create filtered views, reports, follow-up tasks, or automations.

Here is the second AI assistant we will use in this guide, called "Application Rank Assistant", and its settings to rank the candidate from 1 - 10.

With our API key generated and both AI assistants set up, we are now ready to create a Flow in Knack.

---

How to Configure: Knack to Open AI assistant flow

Overview:

In my Knack app, titled "AI HR Recruitment App", I’ve created a Job Applications table with fields commonly found in a job application form — such as Name, Email, Years of Relevant Experience, Education Level, and open-ended questions like "Tell us about yourself" and "Why are you interested in this role?" These fields are used to collect applicant data through a form.

Additionally, I’ve included two custom fields: OpenAI Candidate Summary and OpenAI Candidate Rank (1–10). These fields are designed to store the AI-generated summary and ranking of each applicant, based on their responses.

Here is my Job Application form in our Knack app builder. You can see all the relevant fields are being used for the job application form input:

This form will be used to collect applicant information, which we’ll then send to OpenAI to automatically generate a summary and a ranking for each candidate.
   
 Here is what our 4-step Flow will look like:

• Step 1: Trigger New Form Submission in Knack: When a candidate fills out and submits the form, it will automatically trigger the Flow.
• Step 2: Action Message to AI Assistant in Open AI: We will send this form's data to our "HR Screener" AI assistant for a detailed summary and analysis.
• Step 3: Action Message to AI Assistant in Open AI: We will then send step 2's AI summary to our "Candidate Rank AI assistant" to rank the candidate from 1 - 10:
• Step 4: Action Update Record in Knack: Afterwards, we will update our Candidates Application record with the AI's summary and rank. This can be used to then trigger other automations or tasks, or be used for filtered record views and reports.

Let's Build our Flow, start by clicking Create Flow From Scratch in our Flows builder section:

STEP 1 TRIGGER: NEW FORM SUBMISSION:

Click Add a Trigger: when this happens in the selected app

Choose Knack as your app, and Form Submission as your Event to trigger this Flow:
In the "From Tables" field, I will choose the Job Applications shown below:

Next, select the form you want to use as the trigger. In my app, the form shown above is called "Add Job Application."

Click "Save" to activate the Flow. From now on, each time a candidate submits an application, the Flow will run automatically.

Next, let's send this data to OpenAI to analyze and rank:

STEP 2 ACTION: SEND A MESSAGE TO AI ASSISTANT (HR SCREENER) TO PERFORM A DETAILED ANALYSIS.

Add a new action:

Choose ChatGPT (OpenAI) for your app and Message to Assistant for the Action you want to perform:

To connect OpenAI in the Connect your ChatGPT(OpenAI) app, click on + Add Linked App:

Here you can paste the API keys you copied from earlier; this will connect your OpenAI account to Knack. Click continue. 

Next, you will be presented with these settings:

• Assistant - ID (required): Select the AI Assistant you’ve configured in OpenAI, or enter the assistant’s ID manually.
• Message to Assistant (required): Enter the text or pass data from previous steps that you want the AI to analyze or respond to.
• Thread ID (optional): Use this to continue a conversation with an existing thread, preserving message history. If left blank, a new thread is created automatically.
• Override Instructions (optional): Customize the assistant’s behavior for the current run by specifying temporary instructions, which will override the default system message.
• Additional Instructions (optional): Append extra instructions to modify the assistant's response for this specific request without changing the default settings

For this example, I will stick with the Assistant - ID and Message to Assistant fields:
   
When you click into the Assistant - ID field, you will see your linked AI Assistants to choose from. In this instance, I want to choose HR Screener, which will provide me with a detailed text summary of the candidate:

The Message to Assistant section is where we provide instructions to the AI, along with the relevant mapped fields.

As shown here, I first typed out the field names manually to give the AI context, and then mapped the values from our Knack form submission in Step 1:

The data will now be sent to OpenAI, which will generate a detailed analysis of the candidate’s application.

Next, let's map the AI Assistant's response to our next Open AI Assistant step to rank the job candidate from 1 - 10.

STEP 3 ACTION: MESSAGE TO AI ASSISTANT (CANDIDATE RANK) TO RANK FROM 1 - 10:

Add a new action, 

Again, choose ChatGPT (OpenAI) as your app and Message to assistant as your Action:

Here we will choose our other AI assistant, Application Rank Assistant, for our "Assistant - ID" field.

We will then map the RESPONSE MESSAGE output from Step 2 into the Messages field:

The AI will review the summary from Step 2 and give it a score from 1 to 10. We’ll then take both the summary and the score from OpenAI and save them to the new Job Application record in Knack.


STEP 4: UPDATED RECORD IN KNACK:

Add another action to your Flow:

For the app, select Knack, and for the action, choose "Update Record." Then, select the Job Applications table as the one you want to update.

Click "Continue." On the next page, you'll see the following settings:

For the required Record ID (required) field, since we're updating the same record that was created in Step 1 (after the form was submitted), map the Record ID from Step 1 — Knack: Form Submitted — like this:

Lastly, map the output from your two AI steps to their corresponding fields.

For the OpenAI Candidate Summary field, I mapped the Response Message from Step 2: Message to Assistant. This provides a detailed summary of the candidate, generated by our AI Assistant (HR Screener).

For the OpenAI Candidate Rank 1 - 10 field, I have mapped the AI assistants (Application Rank Assistant) Response Message from Step 3:

Click "Save" — your Flow is now fully configured.

You can now start the Flow to begin capturing job applications for review. From this point on, every time an applicant submits the application form in Knack:

• Step 1: A new record is created in Knack
• Steps 2 & 3: The data is sent to OpenAI for analysis and scoring
• Step 4: The AI-generated summary and score are added back to the Job Applications table in Knack for easy review
  
  

Let’s test it out!
Submit a job application to trigger the Flow and watch it in action. 🚀

Give it a moment for the AI steps to process the data.

And there it is — our job application has been successfully submitted!
A new record has been created and updated with the applicant’s answers, OpenAI’s analysis, and a rank between 1 and 10.

Now imagine this at scale — hundreds or even thousands of applications, all automatically summarized and ranked by AI for your HR team to review and take action on, or to trigger additional automations.

For example, you could create a view in Knack sorted by AI Rank, from highest to lowest, making it easy to prioritize top candidates at a glance.

Lastly, you can build on this Flow by adding even more action steps to automate your recruitment process.

For example, you could:

• Automatically send an email and schedule an interview if a candidate's rank is greater than 8
• Set up scheduled tasks in Knack to send follow-ups or update records based on specific criteria
• Add action links to your Job Application views to manually approve candidates or send emails directly from the interface

Here are some powerful next-step actions to make your HR Flow even more effective:

• 📧 Send email notifications to HR or candidates
• 📅 Create interview tasks in your project management tool
• 📇 Add candidates to a CRM or talent pipeline
• 🗓 Schedule interviews with calendar invites
• 📊 Update reports or dashboards automatically
• 🧭 Route candidates based on their AI score
• 🔔 Send alerts via SMS or Slack
  
  

---

Versatile AI + Flows Use Cases Across Industries

Discover how AI assistants and Flows can transform workflows in a wide range of industries:

• Education: Automate student application reviews and generate personalized study plans.
• Healthcare: Prioritize patient intake, automate follow-up reminders, and summarize medical histories.
• Business Operations: Streamline invoice approvals, analyze customer feedback, and automate vendor onboarding.
• Customer Support: Use AI to triage support tickets, suggest solutions, and escalate urgent issues.
• Sales & Marketing: Score leads, generate personalized outreach messages, and track campaign performance.
• Human Resources: Screen candidates, automate onboarding checklists, and monitor employee satisfaction.
• Nonprofits: Qualify volunteers, track donations, and manage event registrations.
• Small & Medium Businesses: Automate order processing, send real-time updates, and analyze sales trends.
• Finance: Automate expense approvals, detect anomalies, and generate financial summaries.
• Legal: Summarize case documents, automate contract reviews, and manage client intake forms.

This wide variety of use cases shows how AI assistants + Flows can make your processes smarter, faster, and easier, no matter your field.

Interested? Try it out yourself — sign up for a free Knack trial and start building your own AI-powered workflows: https://dashboard.knack.com/sign-up