I use Vertex AI Pipeline to predict delivery days for my ecommerce. What I appreciate is having everything as a one-stop shop, which has made my life easier as someone semi-technical. Since I have a very extensive relationship with Google products, it works well for me. I like the Google Cloud Platform and feel that the UI and general design philosophy make it easier for me to use, even without a heavy data science background. The integrated development, configuration, licensing, and integration in one spot is really convenient. Review collected by and hosted on G2.com.

I find the analytics and accuracy lacking, with a lot of hallucination happening, especially during my first trials with bigger data models. It's crucial to have extremely precise data to get better output. The user interface was definitely a challenge for me initially. Even though I like Google's design philosophy, I kept getting lost and had to frequently search online to figure things out. Moving from A to Z wasn't intuitive. Review collected by and hosted on G2.com.

It holds a conversation across multiple turns. Remember last conversation context and reply on the basis of last query. Easy to use and quick response Review collected by and hosted on G2.com.

Sometimes it become slower but that is manageable because mostly it respond in time only Review collected by and hosted on G2.com.

What stands out most about Vertex AI is how it unifies the entire ML lifecycle in one managed environment. Data prep, training, hyperparameter tuning, model registry, deployment, monitoring, and now foundation model access through Gemini are all integrated. The tight coupling with BigQuery and Cloud Storage reduces data friction significantly.

I also appreciate the managed infrastructure. You get scalable training on GPUs and TPUs without wrestling with low level provisioning. Experiment tracking, model versioning, and endpoint autoscaling are built in, which makes it production friendly. For teams deploying LLM powered apps, the generative AI APIs and evaluation tooling are particularly strong. Review collected by and hosted on G2.com.

The learning curve can be steep. There are many moving parts across projects, service accounts, IAM roles, networking, and quotas. For smaller teams or solo developers, initial setup can feel heavy.

Cost visibility can also be challenging. Training jobs, prediction endpoints, storage, logging, and networking all accumulate charges separately. Without strong monitoring, it is easy to overspend. The UI is powerful but sometimes inconsistent across different services, and debugging distributed training jobs is not always straightforward. Review collected by and hosted on G2.com.

With the Gemini Enterprise Platform, I’ve gotten better results for coding, UI/UX design, image generation, and video generation. Overall, it’s helped me produce stronger outputs across these areas. Review collected by and hosted on G2.com.

It only generates 10-second videos, and there’s a limit of five videos per day. Review collected by and hosted on G2.com.

I like vertex ai agent builder turns complex agent design, grouping, and deployment into a single, productive workflow that saves development time and reduces operational risk. Review collected by and hosted on G2.com.

The main dislikes are cost, google-cloud lock-in, and complexity for advanced customization. Review collected by and hosted on G2.com.

vertex AI Agent Builder makes it easy to build AI agents quickly ,integrates well with google cloud, and delivers reliable, scalable performance. Review collected by and hosted on G2.com.

The pricing can be high for larger deployments, and some advanced feature have a learning curve for new users. Review collected by and hosted on G2.com.

End-to-end tooling: It brings together retrieval, orchestration, evaluation, and deployment instead of making you stitch everything together yourself. Review collected by and hosted on G2.com.

Abstraction can be limiting: The convenience is great early on, but once you want custom orchestration, nuanced tool use, or unusual workflows, the platform can feel restrictive Review collected by and hosted on G2.com.

I like how quickly I can build and test agents without spending too much time on setup and it makes development much smoother. Review collected by and hosted on G2.com.

The setup can feel a bit overwhelming at first, especially if you are new to the Google Cloud ecosystem. Review collected by and hosted on G2.com.

The all in one ecosystem integration, we can make data pipelines as well as train data models in the same system without having to move them to another one. and we can also get access to other open source models along with google's gemini and foundational models Review collected by and hosted on G2.com.

If compute is not configured correctly, it can tun endlessly incurring hight costs, also its billing is very complex. And also, for my individual projects, it is very costly. Review collected by and hosted on G2.com.

What I like most about Vertex AI is how it brings the entire machine learning lifecycle into one well-organized platform. It simplifies everything from data preparation and model training to deployment and monitoring, which makes even complex ML workflows easier to manage. The tight integration with Google Cloud services adds real value, especially when it comes to scalability, security, and performance. Overall, Vertex AI strikes a strong balance between the flexibility advanced users need and the ease of use teams want when building reliable, production-ready machine learning without a lot of extra overhead. Review collected by and hosted on G2.com.

What I don’t like about Vertex AI is that it can feel overwhelming at the beginning, especially for users who are new to Google Cloud or ML platforms. The learning curve is steep, and it takes time to understand how all the services, permissions, and pricing pieces fit together. The documentation can also feel a bit fragmented, which makes it harder to find clear, end-to-end guidance for specific use cases. On top of that, costs aren’t always easy to predict without close monitoring, which can be challenging for smaller teams or budget-conscious projects. Review collected by and hosted on G2.com.