You've made it to the end of this course on analytics and AI. Let's recap what you've learned. We started the course by making the case for machine learning transformative potential. Machine learning has already helped many businesses scale, and be more efficient. Machine learning can likely be applied to your data to help you get actionable insights for your businesses. We then did a high level overview of the GCP ML ecosystem. Next, we explained that while unstructured data is omnipresent in a business, it's often very difficult to extract value from unstructured data. We then introduced several options on GCP, for doing just this, such as the Cloud Vision, and Cloud NLP APIs. These options use Google's models, and Google's data. So they're geared towards clients who either don't have a need for customized ML models, or who simply do not have the resources to develop them. We then introduced AI Platform Notebooks as a perfect tool for prototyping machine learning pipelines. AI Platform Notebooks allow engineers to collaborate on machine learning in an environment designed specifically, for ML development. Not to mention Notebooks integrate nicely with other GCP services. We mentioned that GCP has a wide range of ML solutions depending on a businesses resource and needs. For those who have in-house ML expertise, we made the case for AI platform which allows you to train customize ML models on your own data. For businesses with no ML expertise, we introduce the perception APIs. For businesses in between, we introduce AutoML as a viable option. Also, we talked about Kubeflow as a tool for putting ML models into production. We also talked about AI Hub, a repository for ML pipelines and models. Remember, don't reinvent the wheel. Look on AI hub first to see if there are any resources that can be used as a starting point for your needs. Next, we discussed how BigQuery can be used for building machine learning models and making predictions. This is a very powerful option for ML since you can build ML models in the same place your data lives. Finally, we concluded the course by going over the range of services on Cloud AutoML. Cloud AutoML trains machine learning models on your data, and Google's models. So there's a level of customization you can achieve that you cannot if you use the perception APIs. Another advantage of using Cloud AutoML is that you can get a customized ML model, without having to do any coding.