Oracle has been upping its AI and machine learning game of late, as the company continues to push its vision of using the technology to shift business’ time away from ‘mundane’ tasks.
The company has weaved AI and ML into a number of its offerings, including its security cloud, its autonomous database and most recently its wider cloud platform, clearly positioning automation as a core focus of the company.
We sat down with Oracle cloud platform VP of product management and strategy Siddhartha Agarwal to discuss how Oracle views AI technology development as well as what it sees as the big challenges for the enterprise space this year.
What do you see as the key technology challenges facing the enterprise for 2018?
There are certainly a number of challenges facing the enterprise space. Firstly, I would say there is a challenge around building serverless applications, where they want to be able to have things running just for the time that they are needed and they don’t have to worry about the infrastructure that it is running on.
The challenge here is that you are effectively doing distributor programming because you have a function that comes up and then goes away and you have to chain together a bunch of different functions. Then there’s the challenge of portability and being able to run these functions on containers without necessarily having to know or manage these containers.
There is also an emerging challenge around blockchain and where the enterprise could use blockchain in their applications. This can be difficult to determine as blockchain is generally for enabling transactions between multiple parties, where not all parties are fully trusted. Although it allows a secure transaction where third-party intermediaries, which introduces significant efficiency.
So enterprises need to ensure that they’re hyperledger fabric, ensure scalability, ensure it is integrated with identity management and maintain performance as the chains get larger. So the issue becomes about how to leverage blockchain without worrying about the infrastructure and management that goes along with the hyperledger implementation.
There is also another trend around IoT. A lot of folks have focused on building the platform for IoT in terms of how to ingest the data, run predictive analytics and integrate the data with backend applications. However, the challenge is that businesses don’t necessarily want to maintain a technology platform and then build extensions and enhancements. They just want to leverage Software-as-a-Service (SaaS) type capabilities where things like asset monitoring and predictive maintenance is visible to them as a SaaS service.
So there are a few challenges that enterprises will have going forward as new technologies develop, and that is just a few that we’re focusing on.
We’ve already been talking a lot about AI and automation in the enterprise space this year. How do you think attitudes to this will develop going forward?
I think the reason why AI or machine learning (ML) is becoming popular comes down to two things. First, there is a significant amount of data now available and in order for machine learning to be successful, you need to have a great deal of data. Second, we now have an increased amount of compute capacity that’s available on-demand and GPUs that can process ML algorithms much faster, to train model or to come up with the heuristics.
When you’re looking at implementing ML, you first have to assess how you would benefit from ML being embedded into the various capabilities that you get from software providers. Thinking about Oracle’s autonomous data warehouse, we are now using machine learning to be able to optimise queries and we actually turned off the ability for some personnel to turn off these queries. So there are numerous advantages - in our experience - of AI and being embedded in different platforms that organisations benefit from.
Do you think the A/NZ region is showing a good responsiveness to AI, or have we got a lot to learn?
No, I think that there are definitely customers in this region that are already leveraging AI in at least the consumption sense. An example is Adelaide-based National Pharmacies, who in the past had built a mobile application to be able to drive loyalty programs and various customer service offerings. They’re now experimenting with chatbots to drive customer service interaction and significantly reduce the wait time and enhance customer responsiveness.
I don’t think it’s about the Australian market not being advanced in AI implementations. What’s more relevant is thinking about business use cases and being able to consume technology without having to be AI experts. Oracle is looking to democratise AI, as opposed to having it in an inaccessible realm only accessible and valuable to data scientists.
What advice would you give to organisations looking to evolve their AI-driven offerings with the plethora of options available in different areas at the moment?
I think firstly I would urge organisations to identify business cases where AI would actually be able to help them. Then, look for solution providers that give you some AI out of the box, so you can actually benefit from it rather than having to build it from scratch. It would be troublesome to start with the attitude of wanting to build an AI or ML practice because there aren’t many people who have the expertise to deliver that, like data scientists or AI experts.
Oracle recently made a substantial announcement around its autonomous platform service, How have you found the response to this so far?
The response from the customer community has been really good. Our vision for the autonomous PaaS is that software should be able to configure, manage, scale and tune itself, while automatically fixing and avoiding problems. Customers are generally experiencing lower cost and increased productivity because they feel they can eliminate a significant amount of human labour.
They are also experiencing increased agility, as traditionally if they wanted to set up a new data mart, they would need to get data into the data mart, get the data mart stood up and then run analytics through it. That process would have taken from 6 - 9 months if those customers were lucky.
Now you can easily get an autonomous data warehouse running where you don’t need the DBA skills to be able to do things like optimise queries and set up and scale the data warehouse. You’re able to consume it, use it and set up a functional data mart with analytics running on top of it within a month.
The autonomous cloud platform also alleviates the issue of patches being available but not applied as once it realises the patch is available, it automatically applies the patch while the application is running, without experiencing significant downtime.
So I think the customer reaction has been extremely positive and we actually think autonomous should be the de facto standard for all PaaS vendors out there.