Google Vertex AI Reviews

My research was around how well they support RAG use cases. Do they have all the components required for building the RAG use case? As in, you need a data storage part, vector storage where you can store your vector embeddings, and the model layer itself.

Vertex comes with inbuilt integration with GCP for data storage. It also supports ChromaDB, which is not provided by Azure.

I'm not sure if I have suggestions for improvement. Based on my comparison between the two, Vertex has various additional functionalities that Azure doesn't provide.

I have been using it for quite some time, approximately two months.

I found Vertex to be quite more stable than Azure.

It is quite scalable. It can easily scale up or down based on user demands.

I haven't personally tried customer service, however, based on reviews, they have good customer support that is available during downtime.

Positive

Before these two, we used H2O AI. It lacks most of the features, and H2O AI pieces are scattered rather than having a unified platform.

I find it quite user-friendly. A beginner can also find it easy to understand. The learning curve is exponential.

They have different pricing models like pay-as-you-go or subscription model, and total cost of ownership. It is comparatively cheaper than Azure.

I would definitely recommend Vertex AI to someone. I compared both platforms over a wide range of factors, and in most areas, Vertex AI is ahead of Azure.

I'd rate the solution nine out of ten.

We were looking for optimization for a machine learning module. Gurobi is an optimization library, but it's quite expensive. We were looking for an alternative to Gurobi and came across Google Vertex AI. We still find that Gurobi works better than Google Vertex AI. Eventually, we dropped out after six months of using Google Vertex AI.

Google Vertex AI is an out-of-the-box and very easy-to-use solution. You don't need an army of data scientists to help you build your model. The solution is really easy to use. If we teach or tell the business people what to select and how to train, they can do it by themselves in three months. That's how easy it is to use Google Vertex AI.

Google Vertex AI is good in machine learning and AI, but it lacks optimization.

I have been using Google Vertex AI for six months.

Google Vertex AI is a stable solution.

The solution's initial setup is pretty straightforward and doesn't need much technical expertise. It has some samples and libraries, and you can quickly do it yourself. Eventually, if you want to do more customization, you will need the help of technical people.

We chose Gurobi over Google Vertex AI because Gurobi's library had 95% of our expected result, but Google Vertex AI had 75% in optimization modeling. Google Vertex AI is good in machine learning and AI, but it lacks optimization, and we needed optimization.

Users who already have Microsoft Office 365 should use Microsoft machine learning and AI. If they don't have Microsoft Office 365, they can use Google Vertex AI.

Overall, I rate Google Vertex AI an eight out of ten.

Our use case involves leveraging it for tasks such as generating document summaries with keyword detection after scanning. Additionally, we employ image augmentation based on image descriptions. For advanced language analytics, we analyze the frequency of specific keywords and business terms within documents, ultimately ranking the documents based on these criteria.

It provides the most valuable external analytics. Whether I'm searching for patterns or conducting keyword-based analytics, I consistently achieve better accuracy with Vertex compared to OpenAI from Azure.

While it employs chat activity for answering queries, the available options might be somewhat limited. I've noticed that using chat activity often presents a broader range of options and insights for a well-constructed question. Improving the knowledge base could be a key aspect for enhancement—expanding the information sources to enhance the generation process.

We have been using it for the past three months.

I find it easy to both scale out and scale up. I don't foresee any issues when the user volumes increase; the Google Platform automatically handles it, which is helpful.

Google's support is impressive. They are highly responsive and go the extra mile to assist users in trying out and ensuring they stand out in the market. I would rate it ten out of ten.

Positive

Previously we were using Azure OpenAI. In certain aspects, Azure outshines Vertex, particularly in total text analytics and, more notably, in multi-language capabilities. However, when it comes to generating content from a robust knowledge base, OpenAI is my go-to because of its extensive knowledge repository. The choice between platforms depends on the specific use case at hand.

The initial setup was easy, especially for someone with a technical background like myself. It didn't require much effort, and the abundance of documentation made the process smooth.

Our current activities involve proof of concept and piloting. Currently, our pilot phase appears to be successful, and we are examining volume metrics to determine the next steps for transitioning into actual production deployment.

As a product company with implementations spanning eight different platforms, we plan to offer this specific feature selectively to enterprise-grade customers based on their needs. I'm confident that there's substantial value addition for end customers using it, leading to a significant return on investment.

Overall, I would rate it eight out of ten.

In my company, I use the Google Cloud Platform solution for Vertex AI Workbench and Vertex AI with Gemini. I also use Vertex AI Agent Builder and the other stuff they offer as a framework to build different kinds of applications.

I won't be able to give you the specific details for the customers, but it is mainly for building the operational application for our company's clients by focusing on GenAI capabilities so that at least we can optimize the work, which usually comes in the agent support.

The most valuable features of the solution are that it is quite flexible, and some of the services are almost low-code, with no-code services, so it gives agents flexibility to build the use cases according to the operational needs. If somebody wants to build something on their own, it is very easy for even the end users. On the contrary, if they want to customize the pipelines and cover the granular details, they can utilize the options available for capturing and controlling every pipeline stage with codes.

The tools' models are pretty good. The multi-model capabilities from Vertex AI with Gemini 1.5 Pro are quite good, but the cost is slightly higher.

I think the technical documentation is not readily available in the tool. From the ecosystem perspective, there is not enough information available, even for partners or developers who would write certain blogs or content explaining how one can utilize the tool's services. If something gets stuck, it is difficult to access. The documentation is not readily accessible.

I have been using Google Vertex AI for a year. My company has a partnership with Google. My organization also operates as an integrator for Google.

It is a stable solution.

The tool's scalability features depend on the provisioning, but the tool is actually quite scalable, and there is no limit. You can start as low as with a single node and one CPU and go all the way up. There is no limit to how much you can scale up since it all depends on the pricing tier and the service configuration you utilize.

I have experience with Microsoft, and I feel that it is much better than Google. Microsoft's service maturity is slightly higher regarding features or capabilities, and it is slightly more mature as an RPA tool than Google.

The product is easy to provision and install.

The time needed to deploy the solution depends on the type of application. It can take as little as three weeks, or it may go on for even three months.

A very small scrum team of two to three resources can actually build the entire use case from start to end.

In general, ROI largely depends on the use case, not the service cost itself. The footprint can be revised based on the operational gains. We always start low, and the infra costs are usually a very small fraction of the resource cost.

I think almost every tool offers a decent discount. In terms of credits or other stuff, every cloud provider provides a good number of incentives to onboard new clients.

Speaking about how Google Vertex AI's automation feature improved our company's ML workflows, I would say that we don't have to do much. Suppose you totally rely on the tool's pre-automated workflows. In that case, it will take away a huge effort that is involved in building retrieval augmented generation pipelines. So you just have to dump the dataset, and automatically, it will be able to fulfill the needs of the pipeline.

Some people would be required for operational support, mainly for fine-tuning and optimizing the application.

Speaking about how the tool helps enhance data management and offers integration capabilities, I would have to say that such features are common for all cloud providers. The native integrations are quite easy, but integrations with third parties are usually tough.

Before choosing Google Vertex AI, people should evaluate all the services and compare the vendors.

I rate the tool a seven to eight out of ten.

We use the solution for the model garden. We leverage both open-source models and Google Vertex AI's embedding models and use them to deploy our machine-learning models.

The most valuable feature we've found is the model garden, which allows us to deploy and use various models through the provided endpoints easily.

The tool's documentation is not good. It is hard. 

I have been working with the product for two years. 

The tool is stable, and I haven't faced any bugs. 

Google Vertex AI is scalable. My company has four to five users. 

The tool's deployment was easy. The Cloud Run feature was tricky. 

The solution's pricing is moderate. 

I would recommend using Google Vertex AI. It's a good product, but it depends on your use case. If your needs align with its capabilities, it's worth considering. I started as a beginner and found it easy to learn and use. I rate it an eight out of ten. 

We use Vertex AI for building machine learning workflows. This encompasses the entire process, from developing the workflow for training models to making predictions. Additionally, it handles the integration of diverse data types, including electronic data interchange (EDI), Salesforce (SFDC), and other formats. This aligns with the information you inquired about during our discussion last week.

We use Google Cloud's Vertex AI platform for our machine learning workflows. Specifically, we leverage the IO branch for EDA data. The AI model registry in Vertex AI is crucial for cataloging and managing various versions of the models we develop. When it comes to deploying models, we rely on Google Cloud's AI Prediction service, seamlessly integrating it into our workflow for real-time predictions or streaming. 

For monitoring and tracking the outcomes of model development, we employ Vertex AI Monitoring, ensuring a comprehensive understanding of the model's performance and results. This integrated approach within Vertex AI provides a unified platform for managing, deploying, and monitoring machine learning models efficiently.

I believe that Vertex AI is a robust platform, but its effectiveness depends significantly on the domain knowledge of the developer using it. While Vertex AI does offer support through the console UI in the Google Cloud environment, it is better suited for technical members who have a deeper understanding of machine learning concepts. 

The platform may be challenging for business process developers (BPDUs) who lack extensive technical knowledge, as it involves intricate customization and handling of numerous parameters. Effectively utilizing Vertex AI requires not only familiarity with machine learning frameworks like TensorFlow or PyTorch but also proficiency in Python programming. The complexity of these requirements might pose challenges for less technically oriented users, making it crucial to have a solid foundation in both machine learning principles and Python coding to extract the full value from Vertex AI.

I have been using Google Vertex AI for two to three months. 

I would rate the stability of Vertex AI at nine. It has proven to be highly reliable for both business and technical aspects. On the business side, Vertex AI empowers users to explore and implement machine learning models effectively, contributing to business growth and data analysis. 

From a technical perspective, the infrastructure is robust, ensuring there is no downtime. The seamless integration with other tools like Cloud Composer and Dataflow allows for comprehensive control over the entire machine learning workflow, creating a cohesive and efficient experience. 

The scalability of Vertex AI is a perfect 10. Leveraging the robust infrastructure of Google has evidently provided us with ample resources to seamlessly run our machine learning processes. The absence of any limitations or constraints allows us to execute our tasks without hindrance, highlighting the efficiency and reliability of the tool.

While the support is generally good, there have been instances where the response time was slower than expected. Since Vertex AI is a managed service, the support team sometimes takes longer to address queries as they need to escalate them to the product team. I believe there is room for improvement in terms of faster resolution of tickets and quicker responses, especially for issues that could be resolved more promptly.

Positive

The initial setup with Vertex AI poses no challenges, especially given the seamless interaction with Google Cloud. However, I acknowledge that the platform involves a plethora of parameters and technical intricacies that must be addressed to configure the environment correctly. It seems like a more straightforward, user-friendly approach to configuring these parameters would be beneficial, ensuring that users can easily and accurately set up the environment without encountering complexities. Simplifying these technical aspects could further enhance the overall user experience with Vertex AI.Our workflow typically begins with the initial step of acquiring the Vertex AI API, a prerequisite for utilizing Vertex AI's capabilities. Subsequently, we proceed with the Vertex AI Workbench, focusing on exploratory data analysis (EDA) to extract relevant features crucial for our machine learning models. Following this, we engage in Extract, Transform, Load (ETL) processes, manipulating and storing data in platforms like BigQuery or Cloud Storage.

The subsequent phase involves coding the machine learning algorithm and encapsulating it into an image bundle. This bundle is then pushed into the Vertex AI training environment, initiating the training job. Once the model is successfully trained, we leverage the Vertex AI Model Registry for efficient model deployment into production. This deployment can take various forms, such as batch predictions or real-time streaming endpoints.

The final stage of our workflow often includes addressing customer requests for model monitoring, a task facilitated by the Vertex AI Monitoring tool. While some clients opt to include this monitoring step, others conclude their workflow at the AI model deployment stage. Overall, our workflow is comprehensive, covering all essential aspects from initial data analysis to deploying fully trained models into production.

Vertex AI offers attractive pricing. With this pricing structure, I can leverage various opportunities to bring value to my business. It's a positive aspect worth considering.

I would rate the overall solution 7 out of 10.