Azure OpenAI Reviews

We can use the solution to implement our tasks and models quickly.

The product must improve its dashboards. I would like to know how many tokens my requests use and how accurate the search is. We are using Python language to find out about it.

I have been using the solution for six months. I am using the latest version of the solution.

I rate the tool’s stability a ten out of ten. I do not have any problems with it.

About 5000 people use the tool in our organization. I rate the tool’s scalability a six out of ten because it doesn’t have automatic installation options.

We have done some POCs with AWS Bedrock and some AI products from Google. We chose Azure OpenAI because we have a preference for the vendor. All our infrastructure is deployed on Azure.

We are maintaining the solution, but the maintenance is not on OpenAI services. We maintain the Python code that we embedded. Our technical team has five people, including two engineers and two data scientists.

The response time of our call centers has improved. The team is using the language models to ask simple questions.

I would recommend the tool to others. Overall, I rate the solution an eight out of ten.

I was freelancing for a company that wanted me to make tutorials on how the platform can be used. So, here are just a few model-building video tutorials I made from the platform. That's pretty much it.

It's very easy and convenient to use compared to others. It has good documentation, and it's very easy to follow. So somebody using it for the first time finds it very convenient.

The solution has a very drag-and-drop environment. Instead of coding something from scratch or understanding any concept in extensive depth before deployment, this is good. Plus, they have an auto dataset, which means you can choose any dataset they have instead of providing your own. So that's also pretty nice.

Maybe Azure OpenAI could provide a few video tutorials, in addition to the documentation. If they want to make it easier for somebody to do it for the very first time, providing video tutorials might be a good idea.

So, I would like to have a tutorial added for new users. 

I have only worked for around a month or so.

I would rate the stability a nine out of ten. It is very stable. 

I would rate the scalability a seven out of ten. 

I took up a course that gave me access to Amazon. But when I compare OpenAI with Google and Amazon because I work with both Google and Amazon, I would put OpenAI, then Google, then Amazon.

So, Azure OpenAI is on top of my list. They've got a very user-friendly platform, so that works best. Amazon is slightly complex. Google provides video tutorials, but somehow Azure has a better UI.

I would rate my experience with the initial setup a seven out of ten, where one is difficult, and ten is easy. 

Deployment was slightly complex for me to understand. So, my senior was working on it, but I did not directly deploy it. The instructions are very clear on how to deploy it, so it is fine, and it doesn't take a lot of time. It hardly takes a few minutes, I think, d depending on the data. If the dataset is very big and if the model is complex, then maybe deployment will take more time. But if it's something very simple and basic, deployment was fine.

I would suggest you should give it a try. Overall, I would rate the solution an eight out of ten. 

The typical use cases include building chatbots for financial document analysis, agents for transaction categorization, and call centre voice identification or conversation analytics.

Two aspects I appreciate are the turnaround time and ease of use. As it's a managed service, the quick turnaround is beneficial, and the simple interface makes it easy to work with. Performance and scalability are also strong points since you can scale as needed. 

We used Azure OpenAI to analyze call center voice data. This helped us better understand customer sentiments and make recommendations.

I have found the tool unreliable in certain use cases. I aim to enhance the system's latency, particularly in responding to calls. Occasionally, calls don't respond, so I want to improve reliability.

I have been working with the product for six months. 

I have issues with Azure OpenAI's stability and reliability. 

The tool's scalability is good. 

I have worked with Amazon AWS but found Azure OpenAPI to be simpler. 

Azure OpenAI's deployment is straightforward. The deployment process takes around half a day to a full day, considering the use case and the end-to-end deployment. It works for around four to eight hours. To deploy the product, typical steps include data analysis, setting up keys for OpenAI, making API calls with the relevant dataset, implementing basic guardrails, and analyzing the final output. These are the basic steps involved in the deployment process.

A project that would have taken three to six months to build was completed in just six weeks with the help of Azure OpenAPI. So, that's our ROI. The biggest value of the service is how quickly you can prototype your use cases. It offers unlimited scalability, and it is easy to find something closer to your country. Plus, it's highly scalable and comparatively cheaper than other solutions.

I rate the overall product an eight out of ten. If you're comfortable with your data being in the cloud and want quick results, Azure OpenAI is a great option. However, I haven't used it in a production environment yet, so I can't comment.

We're implementing an assistant using Azure OpenAI. The challenge is grounding OpenAI responses to our specific data. 

We can only offer users basic querying, like for documents they're stuck on. It handles the request. It's primarily the question-answering feature.

It's very powerful. It allows users to query our documents using natural language and receive answers in the same way. This makes our product information much more accessible than traditional keyword-based search. 

It's focused on information retrieval and question-answering, which suits our needs perfectly. It is more like a natural language query tool we leverage.

We use Azure OpenAI alongside Azure Cognitive Search. These are both new services we've deployed. There's a process where we need to ask Microsoft to create private endpoints to link OpenAI to Azure as a connectivity service.

Since we don't train the model on our data, it's a struggle to ensure OpenAI answers questions exclusively from our data. During user testing, we found ways to make the system provide answers from outside sources. 

As a governance department, accuracy and control are crucial. We're trying to tune the system to stick with our content, but it's an ongoing challenge. 

We've been working on fine-tuning prompts and parameters for about four weeks now.

I've been using Azure OpenAI as a creative source for the past six months.

We've noticed some issues with scaling. It takes time for the service to adapt when we increase the load. We're still in the pre-production phase, and we're seeing this even during testing. 

Also, there's limited capacity in our region (Canada East), which makes it difficult to accommodate the expected load. We've submitted capacity increase requests, but we're not sure if they'll be approved.

The main challenge we've faced is around capacity. Even after running extensive load tests, we don't have sufficient capacity to handle our projected volume.

We have a consultant from Microsoft working with us. They've been very helpful.

However, they're very busy. We could use more of their time if they were available. But they're very competent and helpful. We just wish we could have more access to their expertise.

Positive

We have an alternative search engine that indexes our document base. We use Azure OpenAI's question-answering feature to query that index, generating answers from relevant documents.

We don't use GPT-4 specifically, nor are we training any models. Our IT group leverages Azure OpenAI for its existing capabilities.

It is our first implementation of this kind.

There are some limitations right now. For our specific use case, where we need a traditional information retrieval system, it's not an ideal fit. 

Azure OpenAI is a question-answering system built on top of information retrieval, and that distinction is important for us. Given our use case, I don't think it's well-suited.

Our management team requires accurate and complete results, with precision that matches our existing keyword search tools. It's difficult to evaluate and prove that Azure OpenAI consistently meets that standard. 

We're still early in our adoption, so the rating could change as we deploy it to a larger audience.

For now, I would rate the solution a five out of ten. 

We are assisting our customers in deploying a commercial universal AI solution aimed at aiding them in researching and managing their internal company policies and regulations. To do this, I've extracted all the relevant documents from the HR department and created conversational interfaces for our clients. These interfaces are integrated into various platforms like Microsoft Teams, allowing everyone within the company to interact with the AI.

Its main use for indexing documents and assembling information is highly effective. Previously, we had to meticulously map out each process and step, essentially creating a chatbot for the task.

The most crucial aspect is the conversational capability, where you can simply ask questions, and it provides answers based on your content and documents, particularly tailored to your specific environment.

We encountered challenges related to question understanding. These instances occur when questions are not phrased precisely, resulting in problematic answers. Microsoft is actively addressing this issue and working diligently on improving it.

I have been working with it for six months now.

We have nearly thirty customers using our system, and I can't recall any instances where they've encountered stability issues.

I would rate its scalability capabilities seven out of ten.

We have a direct connection with all the technical support staff in the support area. I would rate it nine out of ten.

Positive

We tried integrating Google in the past, but it didn't proceed as planned so we just stopped it.

The initial setup was straightforward.

The pricing is acceptable, and it's delivering good value for the results and outcomes we need.

My advice is to pay close attention to the content's quality before indexing it within OpenAI. If the documents provided lack good quality, they'll end up with incorrect answers. This is particularly important because the initial setup is not inexpensive and it involves significant investments. Overall, I would rate it nine out of ten.

I use the solution to summarize very long documents, question and answer, build a chatbot, and extract keywords. I build many applications on top of Azure OpenAI.

The models are very intelligent. You just have to write accurate prompts according to your requirements, and the solution gives very good results. You don't need any training data, you don't need to set up your environment completely, or you don't need computational resources. You just pass the prompt with your requirements and get a response.

We had some bad experiences with the solution. We have to send our data to the Azure OpenAI cloud, which they use for training. They say they currently don't use our data for training, but you still have to compromise some secrecy.

The solution's response is a bit slow sometimes. When I use GPT-4, it takes around three to five seconds to generate 100 tokens or a small answer. Many other services perform very well compared to GPT-4. Right now, the issue with GPT-4 is slow response or latency time.

I have been using Azure OpenAI for two years. 

The solution is mostly stable, but there are also some downtimes. It was down a month ago when we were releasing our product. So, we had to wait for the OpenAI servers to work before we could deploy or launch our product. The solution experiences downtime on rare occasions, but it is almost always very stable.

Around 15 to 20 people use OpenAI for production purposes or our applications. Almost everyone uses ChatGPT for daily tasks or generic purposes like coding, text generation, and getting any idea about new products. Around 15 to 20 people use the restricted models, which can be accessed through API.

The solution's initial setup is very easy. You just have to pass the prompt and then hit its API. Integrating the Azure OpenAI models into your application is very easy.

Azure OpenAI is a bit more expensive than other services. Many cloud services and Anthropic AI are cheaper than OpenAI. Many open-source models and API services are also relatively cheap to Azure OpenAI.

We are using both servers. I use Azure OpenAI on our on-premises server. Since OpenAI is a cloud service, we cannot download the Azure OpenAI models on our server. So, we have to use their cloud through the API.

Currently, there is a lot of competition in the LLM area. The person who tries to start with LLMs must try different services, including Azure OpenAI. They should start with the cheapest model, which is GPT-3. They should stick to that model if it works for them and responds well to their requirements. They can also try other cheap API services that respond more accurately to their requirements. I would suggest trying different models or API services to start with.

When you start with the application, it is initially very easy to learn. You just have to write a prompt in simple English language to get the output from the solution. You can write anything; the prompt and your model will yield good results. You can start easily with the solution, but learning more advanced features of prompt engineering will take some time.

Overall, I rate the solution an eight out of ten.

One of the tasks for which I found the use of Azure OpenAI to be useful for my business is related to the area of annotations in images.

Azure OpenAI is not an optimized tool yet, making it one of its shortcomings where improvements are required.

I would like Azure Open AI to provide more integrations with other platforms.

The cost of the product should be lowered.

I have been using Azure OpenAI for six to seven months.

It is a stable solution.

The scalability part of the product depends on whether you have declared the product on an on-premises model and what kind of configurations you are keeping with your back-end servers. I cannot talk about the product's scalability since the tool has more areas like outcomes, precision, and accuracy.

Conversational AI is used across hospitals. The hospital runs Azure OpenAI for EMRs. Businesses have started using AI components for various applications.

The technical support part is documented, and my business works together with Azure OpenAI.

The technical support required by our business depends on the algorithms and the models being developed, which is not what Azure OpenAI provides. It basically lies with the user to solve a problem.

My company works not only with Azure OpenAI but with foundation models, too.

The product's initial setup phase was pretty easy. Installation is not an issue in the tool, but achieving the outcomes matters to our company, which is dependent on algorithms, models, and how much data you use to train your models.

The solution is deployed majorly on the cloud and then on an on-premises model.

The steps that can be deployed in Azure OpenAI include areas like integration with your applications.

Accessibility from your applications and browser is required to deploy the product.

My company has a team of several solution providers who work together. My company has partnered with some of the startups in our ecosystems, so they work with us.

There are around 30 to 40 percent cost-saving outcomes in our company from the use of the solution.

According to the negotiations taking place and the contract, there is a need to make either monthly or yearly payments to use the solution.

With Azure OpenAI, there are a number of alignments that my business is into.

My company works with Azure OpenAI and our own private LLMs.

Though Azure OpenAI is not optimized, it is one of the best when it comes to text generation.

Azure OpenAI is regarded as a foundation model on which our company plans to use our private LLMs.

The natural language understanding capability of Azure OpenAI has improved our company's data analysis since we use the product's integration capabilities for areas like translations and conversational AI.

I recommend the solution to those who plan to use it, but there are also other products that are available on the market.

I rate the overall tool a nine out of ten.

When I write documentation, I use the solution for translating anything and researching.

Azure OpenAI is useful for benchmarking products. We can search for all metrics with AI tools and compare our products with other products.

The solution’s stability could be improved. Azure OpenAI should use more specific sources like academic articles because I can't find the source.

I have been using Azure OpenAI for more than a year.

I rate the solution a nine out of ten for stability.

I rate the solution’s scalability a seven out of ten.

The solution’s initial setup is very easy.

I can deploy the solution by myself in a few seconds.

We pay a licensing fee for Azure OpenAI. The solution's pricing is normal worldwide but expensive in Turkey because Turkey's currency is different.

Azure OpenAI provides me with a lot of benefits. It's quick and saves me time.

Overall, I rate the solution an eight out of ten.