Data management shifted to being powered by AI and ML, and they need data and algorithms - so Oracle has brought them together "to power an outcome for customers."
One trend is the lakehouse – a replacement for data warehouses that is capable of bringing all the different types of data together in one place.
Oracle's distinctive approach to the lakehouse is to provide a converged database rather than trying to combine multiple data repositories.
This way, customers benefit from Oracle's single database engine with autonomous capabilities, bringing all sorts of data into one place that can be accessed via traditional and cloud APIs, using whichever programming language you favour. It can also be combined with Oracle's expertise in industries such as banking (eg, loan application processing), Chelilah observes.
An important point is that this approach bypasses the problems involved in connecting multiple data models, keeping all of the data secure, and staying on top of governance in such a mixed environment.
Most customers are implementing lakehouses in the cloud, he says, and Oracle Cloud Infrastructure is available in 34 regions, with plans to increase that to 44. The company also offers customers the option of a dedicated, on-premises region for applications where latency is an issue (eg, factories) or where data must stay on-premises for governance reasons.
Another trend is the adoption of MLOps (machine learning operations). Turning data into insights for multiple stakeholders who need to interact with the data is a lot simpler with Oracle, he says.
Again, Oracle's approach provides comprehensive, secure and managed data services that are accessible from old and new systems. That includes vertical applications from Oracle and other providers, so, for example, Oracle ERP can be combined with IoT data from the factory floor to gain additional insights, or , or retail POS systems can be linked with sources of weather or traffic data (among others) to generate relevant and appropriate offers.
Chelilah cites research suggesting 40% of organisations that try to apply AI fail to get any business benefits, and that 87% of data science projects fail to go into production.
This approach, he says, helps customers "leapfrog the plumbing and use [the technology] for business outcomes."
The Children's Medical Research Institute has begun adopting this approach, he says, and is extending its pipeline with some of Oracle's new services, which include OCI Language for text analysis, OCI Automatic Speech Recognition (which works in real time), OCI Vision (extract text or visual information from images, eg for document analysis or to identify defective products), OCI Forecasting (for processing time series data), and OCI Data Labeling (to help train models).
Even where an organisation currently uses tools from other companies, Oracle can still accelerate projects by removing the choke points.
Customers who are starting afresh typically don't know where to begin. So Oracle will go through a rapid prototyping process that connects to existing data sources and models.
For example, Oracle has used its Apex low-code development platform to create a home loan processing system that took advantage of the customer's existing ML model for loan approval. Straight out of the box, it reveals the model attributes that contribute most to approvals. The model can thus be monitored over time, allowing data scientists to be called in when it needs attention.
Importantly, Oracle's approach means that the pipeline isn't static or rigid (eg, customers can import a different library if they see fit), and the time to value is reduced because it allows parallel development.