Connecting to ChatGPT on the Megaladata Platform

Artificial Intelligence (AI) is increasingly being integrated into business analytics and data processing. For some tasks, AI is not just a trend but a practical tool. It helps automate routine operations and improve the accuracy of analytical solutions. In this context, AI primarily refers to Large Language Models (LLMs).

In practice, businesses often need more than just "chatting" with a model. They require integration into existing processes, such as implementing intelligent search, automatic text classification and enrichment, comment analysis, or document processing. Automation is key. Requests to the language model should be generated by the system, executed programmatically, and return results directly to the analytical workflow, without manual copying and ideally without user intervention.

                             Megaladata and ChatGPT interaction

Megaladata workflow description

Download the workflow

The file chatgpt_api.mgp contains a ready-to-use connector for the ChatGPT API, complete with example requests.

                                                                Workflow appearance

Important notes:

1. When working with external services and APIs, consider infrastructure and network restrictions. Direct API access may be unavailable due to network environment or access policy limitations. IT specialists typically resolve this using standard tools for secure and stable connections.
2. ChatGPT API access is not free. You need a registered account and a paid subscription. The available processing models (LLM versions) depend on your subscription.
3. Evaluate the token volume of LLM responses. Larger responses may increase the cost of interacting with OpenAI’s servers.

Request structure

Requests in JSON format are submitted to the input port "Table" and follow this structure:

{"model": "gpt-4", "input": "What is Megaladata?"}

In this example, the prompt for the language model is "What is Low-code?". This is the minimal parameter set required for the Workflow to function. For a full list of request parameters, refer to the official ChatGPT API documentation.

The input port for variables contains parameters to configure the connector.

                                                                                     Variable configuration

The table below shows the query parameter settings.

Output configuration

The output table includes:

• Response: Server response in JSON format.
• HttpStatusCode: HTTP response code.
• ErrorDesc: Error description (if any).
• Request: The submitted request.

                                                                                   Output column configuration

The server responds in JSON format.

JSON response from the server

In this example, the LLM response is:

"Low-code is a software development approach that allows developers to design and build applications with little to no hand-coding. Instead, it makes use of visual interfaces, pre-built templates, drag-and-drop blocks, and pre-defined code snippets. The goal of low-code is to accelerate the software development and delivery process, making it faster and more accessible, even for people with limited programming knowledge."."

The rest of the JSON text contains technical details: request descriptions, conditions, errors, etc.

Step-by-step interaction

1. Open the Workflow file chatgpt_api.mgp.
2. Prepare a JSON request with the required structure, as outlined in the documentation.
3. Submit the request to the Table input port.
4. Specify the necessary parameters in the input port for variables.
5. Run the ChatGPT API submodel.
6. Open the output port and verify the response.

Try Megaladata

Using the ChatGPT API connector, Megaladata lets you integrate a large external language model directly into workflows. This enables automatic text processing, intelligent search, description generation, data classification, and meaning extraction all within a single platform.

The solution is not limited to one model. Megaladata workflows can also adapt to work with other LLMs, offering flexibility to select and combine models based on specific business needs.

To evaluate the platform’s functionality, download the free community version of Megaladata here. You can also request a trial server version for your organization by filling out this form.

Integrating LLMs into Megaladata is a practical way to harness the power of modern language models in business analytics.

More on AI in Megaladata:

• AI Case Studies with Megaladata
• AI Code Generation for the Javascript component in Megaladata
• 5 Practical Ways to Use AI in Data Management