 Hello, I'm Mike Dunn with Geophysical Insights, and I'm delighted to share some information about the direction of our machine learning technology and the open subsurface data universe. Before we talk about OSDU as an enabling platform, I'll briefly describe the Paradise AI Workbench for Geoscience Interpretation context. This combines data analytics along with machine learning and deep learning applications. These icons represent the capabilities available to the geoscientist. From left to right, we see the seismic attribute generation icon, and we have one of the largest libraries in the industry, color blending to prepare the data for analysis, attribute selection to identify the best set of attributes, multi-attribute classification, and the associated 2D color map for stratigraphic analysis, geo-body detection to estimate reserves, well-logged visualization to optimize the machine learning results, seismic facies classification and fault detection, which uses deep learning. These technologies are all enabled by GPU computing. We have an easy connector to patrol, and we also deliver the Paradise scripting language, which is used mainly by researchers. The overall takeaway from these applications is powerful and straightforward. More can be seen in the seismic and well data by machine learning than traditional attribute-based analysis alone. And we find that the combination of machine learning applications, which we call machine learning orchestration, refines the final results. If we look at the current state of the many producers who have adopted OSDU, the geoscience interpretation technologies and operating companies are in a state of transition from standard licenses to SAS models to full OSDU operation. In Paradise, we have worked with AWS to develop an OSDU connector, enabling Paradise to run on an OSDU architecture and take advantage of OSDU communications among the various interpretation platforms. Companies using OSDU can realize value from Paradise machine learning applications today, even as we migrate Paradise to a fuller OSDU implementation of the product. Here's a more detailed depiction of how the Paradise OSDU connector is positioned in the IT landscape. The slide is a graphical representation of the Paradise OSDU environment today. The slide illustrates that a user has the ability to log in, search, and push and pull data from the OSDU environment through Paradise. I will share a few examples of the capabilities of Paradise on OSDU beginning with an AI based fault detection, the topic of great interest to the geoscience community. This is a short movie showing highlights of fault detection in Paradise that conveys a good example of machine learning orchestration. This slide depicts the same seismic line with three different data displays. The first is the amplitude data, the second is the auto-pick CNN faults with instantaneous attributes, and the third are the fault neurons with all the other neurons in the display turned off via the 2D color map at the bottom. Now that we have faults as just another attribute, we can create fault geobodies which can be directly input into antracking in patrol. Now we see the combined effect of using the fault, volume, and instantaneous attributes classified in a self-organizing map. The results are both faults and potential reservoirs displayed through sound classification and using the 2D color map. Such a view enables us to see the compartmentalization of hydrocarbons by seeing both a stratigraphic analysis and the faults together in a single set of classifications. To test the validity of the interpretation, the classified results are compared to well logs in an easy to use and convenient panel display. Machine learning geobodies were generated from the SOM classification results shown on the previous slide. The geobodies identify specific portions of the reservoirs. Each geobody is comprised of bins at the sample level and volume metrics can be calculated on geobodies given a few petrophysical parameters. This gives interpreters an initial sense of the size of the potential hydrocarbon bearing reservoirs. Now that the geobodies are color coded and isolated to key reservoirs, we are able to compare the SOM machine learning results directly to the well control. Thank you for listening to our presentation.