AI Content Generation

How to incorporate flexible and enriched AI prompts into your Summit models

🎓

In this tutorial, you'll learn how to ...

❗️

Use the Vault

To keep your API keys to OpenAI and other services secure and encrypted, use the Vault feature of your Summit account. This allows you to then pass in your key securely without entering it as text, like so:

=Gpt("gpt-4o", "{{ MY_OPENAI_KEY }}")

ℹ️

Model Template

The model created in this video can be found in its entirety by clicking the "Augmented Content Generation" quick start.

Video Walkthrough

AI Content Generation Tutorial

This tutorial demonstrates how to use AI for no-code content generation using Summit. It covers setting up API keys, variables, prompt enrichment with searches, and making the model flexible for dynamic content generation through APIs.

Step 1: Access Summit

Navigate to the Summit interface and locate the 'augmented content generation' quick start option on the top left.

Step 2: API Configuration

Click on the GPT step to specify the AI model and to enter your OpenAI API key. Use a variable within curly braces to represent your API key in the configuration.

Step 3: Vault Setup

Go to your 'vault' by clicking on your avatar and then 'vault'. Store the OpenAI API key here for encrypted safekeeping. Ensure the name of the key matches the variable used in the model step.

Step 4: Create Prompt

Define a prompt as an object in Summit. This object will be passed to the AI model. Set a 'cache duration' to specify how frequently the AI can update its response.

Step 5: Enrich Prompt

To make the prompt more specific, use the search feature in Summit to query real-time information from the web and pass it to the prompt. For example, search for the best restaurants in a particular city to enrich the content the AI will generate.

Step 6: Run Model

Execute the model to generate a response using the prompt and search data. This creates content, such as a tweet-length summary, that includes information from the search.

Step 7: Customize Location

Modify the prompt and search to work for any location by turning specific city names into a generic 'location' variable. Also, disable auto-run to prevent unnecessary API calls during these changes.

Step 8: Link Events

Ensure that events (like setting a location) precede searches and prompts in the workflow by connecting them with directed arrows. This sets the sequence for data flow and processing within the model.

Step 9: Test & Publish

Run the updated model to test if it now creates content dynamically based on the new 'location' variable. If successful, set the model live and publish it, enabling the model's API usage for dynamic queries.

General Notes

Variable Sync

When introducing variables, make sure their references are consistent throughout the model to ensure proper data binding and dynamic responses.

Auto-Run Option

Consider using the 'auto-run' feature judiciously during development to prevent unnecessary API usage and potential costs.

API Parameters

When the model is set live, the key variable(s) used in prompt customization become parameters of the model's API, enabling integration with other no-code tools and systems for wider applications.