From Completion to Guidance: Level Up Your AI Interactions with OpenAI Assistants

Introduction

Navigating the vast world of AWS can be overwhelming, especially for newcomers. That’s why we created Cloudwalla – a platform to streamline your cloud management. Think of it as your helpful guide! Cloudwalla’s suggestion feature is like having an AWS expert at your fingertips. Simply describe what you need, and it generates tailored infrastructure recommendations.

The Challenges We Faced

Cloudwalla initially used OpenAI’s completion API to get suggestions. However, this approach needed a lot of specific details in each request. Since the API didn’t maintain a long-term understanding of the project, instructions had to be very precise. This made the process slow and increased the chance of mistakes. We realized Cloudwalla needed a simpler, more flexible way to get suggestions.

Our Initiative

To address our challenges, we replaced OpenAI completion APIs with OpenAI Assistant APIs in the Cloudwalla platform. This streamlined the complexity of interacting with OpenAI models, creating a more intuitive and flexible communication infrastructure.Cloudwalla users can now express their queries directly in text form, simplifying the query process and increasing its efficiency. For instance, when a user needs help in choosing the configuration of AWS EC2, they can simply ask Cloudwalla. This sets the tone for more conversational experience.

The Results Speak for Themselves
 Following the integration of OpenAI Assistant, Cloudwalla saw notable improvements. Not only did the assistant accelerate interaction speeds, but it also reduced the likelihood of errors due to reduced manual instructions. A comparison of the interaction payloads before and after the Assistant’s integration showcases this simplification
 
Before:
{
  "model": "gpt-3.5-turbo-0613",
  "messages": [
    {
      "role": "system",
      "content": "You are a helpful assistant and having"+
      "20 years of consulting experience for AWS ec2 instance."
    },
    {
      "role": "user",
      "content": "I need assistance in choosing the right"+
      "aws ec2 instance type for me."
    },
    {
      "role": "user",
      "content": "Requirements are:"+
      "Application Type: {0}, region: {1} requires continuous availability."
    },
    {
      "role": "system",
      "content": "You have to give response which is very"+
      "cost effective. The cost of an instance should be as low as possible"
    },
    {
      "role": "system",
      "content": "You have to fulfill the given requirements"
    },
    {
      "role": "system",
      "content": "You have to strictly follow the requirements"
    },
    {
      "role": "system",
      "content": "Follow the given format for response and"+
      "do not recommend user to check AWS website"
    },
    .
    .
    .
    .
    "another 10 - 20 lines"
  ]
}

After :

{
"assistant_id": ${assistant_id},
"thread": {
"messages": [
{"role": "user", "content": "application for 100 users in ap-south-2 region"}
]
}
}
In Conclusion

OpenAI Assistant has improved how we interact with LLM. Efficient interactions are possible by replacing the detailed JSON payloads with direct and organic queries.

A Boon for SaaS Multi-Tenant Applications : The introduction of OpenAI Assistant brings significant benefits to SaaS multi-tenant architectures where multiple customers share a single instance of a software application. The assistant can provide dynamic, personalized responses based on tenant-specific data and context. This greatly enhances application usability and customer experience.

Scroll to Top