After a few very busy days in Las Vegas, Oracle AI World 2026 has wrapped up. As you would expect, AI was everywhere. Agentic applications, autonomous operations, AI infrastructure and Oracle's vision for bringing AI closer to enterprise data dominated the keynotes and exhibition floor.
But while AI was the headline, some of the most interesting conversations I had during the week were much more familiar: databases, resilience, disaster recovery, licensing and how customers can get more from the Oracle environments they already have.
And, as is often the case at events like this, some of the most valuable moments happened away from the keynote stage.
Oracle used AI World to reinforce just how far it intends to push availability and resilience.
We saw continued investment in Exadata and Autonomous Database, more automation around maintenance, support for local and remote standby strategies, and extremely aggressive recovery objectives through Oracle's latest Maximum Availability Architecture.
The direction is clear. Customers increasingly expect: less downtime; faster recovery; less potential data loss; more automation; and greater confidence that their systems will remain available when something goes wrong.
Those expectations aren't limited to organisations running Oracle's biggest and most sophisticated architectures. We hear the same questions from customers running Oracle Standard Edition. How quickly can we recover? How much data could we lose? And do we really need to move to Enterprise Edition to achieve the resilience the business expects?
Those questions became a recurring theme in our conversations throughout AI World.
One of my favourite examples came from a conversation with Mario from Multicomputos, who heads up their Oracle practice.
Multicomputos has been a Dbvisit partner, but until our outreach around AI World we hadn't connected with Mario directly. Our invitation was forwarded to him; he came over to meet us at the booth, and we started talking about Oracle Standard Edition and disaster recovery.
He hadn't previously come across our new Zero Data Loss capability, or had much exposure to StandbyMP. That changed pretty quickly once we started discussing what it could mean for their customers.
The conversation moved naturally from the technology to the opportunity within their existing Oracle Standard Edition install base. In particular, we discussed where Standard Edition + StandbyMP + Zero Data Loss could give customers another option when they need very low RPO but don't otherwise require Enterprise Edition.
We've now agreed to continue the conversation with a deeper technical session and demo. For me, that's exactly why being at an event like AI World matters. A relationship that already existed became a completely different conversation simply because we connected with the right person and had something new to talk about.
I also caught up with Bill Rehm from GSI, one of the first people I sold to when I joined Dbvisit. We have known each other for years, but he hadn't yet looked closely at Zero Data Loss. Once we talked through how it works and where it fits, he immediately identified an end customer who could potentially benefit from it.
That's a useful reminder for all of us working with Oracle customers. You don't always need to find a brand-new account. Sometimes the opportunity is already sitting inside an existing relationship. Customer environments change. Business requirements change. Technology changes.
A customer who was perfectly happy with a five- or ten-minute RPO several years ago may now be running a much more critical workload. A Standard Edition environment that once supported a relatively modest application may now be processing transactions the business simply cannot afford to recreate after a failure. The important thing is to keep having the conversation.
AI World was also a great opportunity for us to introduce StandbyMP Zero Data Loss (ZDL) more widely. Traditionally, Standard Edition disaster recovery has involved a compromise around the recovery point. StandbyMP's proven archive-log replication provides a physical standby database, but there is a period between a transaction being committed and the corresponding archive log reaching the standby, typically around five minutes of data loss.
For many organisations, that remains entirely acceptable. For others, it doesn't. ZDL adds continuous redo streaming alongside the existing archive-log replication process, allowing suitable Oracle Standard Edition environments to target a sub-second RPO. That is a meaningful change.
It means organisations no longer necessarily have to look at Enterprise Edition and Data Guard purely because the business wants a much tighter recovery point. And for partners, it creates a very natural reason to go back to customers and ask:
“What does losing five minutes of this workload actually mean to your business?”
That should always be the starting point. If five minutes of data loss is acceptable and easily recoverable, then there may be no need to change anything. But the decision is not only about the value of the data itself. There is also the DBA time spent investigating and reconciling missing transactions, the disruption to users, the stress of managing a recovery when you know there is a data gap, and the confidence that comes from knowing the standby is as current as possible.
Those softer benefits are harder to put into an RPO calculation, but they matter. Less manual recovery work, fewer difficult conversations after an incident and greater peace of mind all improve the day-to-day experience of running a critical database environment.
And where the additional cost of adding ZDL is relatively small compared with the overall DR investment, that can change the conversation from “Do we absolutely need sub-second RPO?” to “Does it make sense to remove this exposure while we have the opportunity?”
Even Oracle wanted to talk Standard Edition
One of the more unexpected conversations of the week was with two Oracle infrastructure sales directors who stopped by the booth. They were genuinely interested in what we were doing with StandbyMP and Oracle Standard Edition. That in itself was encouraging. What was even more interesting was their willingness to discuss Standard Edition customers where we might be able to work together. I hadn't necessarily expected that conversation going into the event, but it reinforced something we have believed for a long time:
There is a place for different technologies and different Oracle editions depending on what the customer actually needs.
Not every workload needs the same architecture.
If Enterprise Edition functionality is required, then Enterprise Edition is the right answer. But when Standard Edition already meets the workload requirements, it makes sense to explore how far that environment can be taken before adding unnecessary cost or complexity. The introduction of sub-second RPO makes that conversation much more interesting than it was even a year ago.
Another highlight was simply seeing so many of our partners face to face. We had partners from across our regions stop by throughout the event, including Cegal, Multicomputos, GSI and others. We also spent time outside the show with partners and friends, including the team from LicenseFortress, with whom we're continuing the conversation around Oracle licensing and architecture.
Those discussions are important because the technology is only one part of the equation. Partners understand their customers' estates. They know which applications are critical, where licensing pressure exists, where infrastructure changes are coming, and where customers are increasingly being asked to improve resilience. That makes them ideally placed to identify where something like ZDL genuinely adds value, and, just as importantly, where it doesn't.
For me, that is a much better conversation than simply trying to sell another product.
It's about helping a customer answer:
What level of resilience do we actually need, and what's the most sensible way to achieve it?
AI World is enormous. There are keynotes, sessions, announcements, demos and more technology than anyone could realistically absorb in a few days. But the thing I came away valuing most was the time spent with people.
Meeting partners we've worked with for years. Connecting with people inside partner organisations we hadn't previously known. Talking to Oracle teams. Hearing customers describe the challenges they're actually dealing with. And spending some proper face-to-face time with the Dbvisit team.
We can hold demos over Teams. We can send white papers. We can talk through architectures on a video call. But there is still something different about standing at a booth with someone, sketching out their environment and hearing them say:
“Actually, I have a customer this could work for.”
That's when a product announcement becomes a real business opportunity.
AI may have dominated the banners, keynotes and product announcements this year. But beneath that, the database conversation remains incredibly relevant. Businesses still care about the fundamentals:
Is my data protected?
Can I recover quickly?
How much could I lose?
Am I paying for more technology than I actually need?
The answers to those questions continue to evolve. For Oracle Standard Edition customers in particular, the arrival of StandbyMP Zero Data Loss changes one of the long-standing assumptions around disaster recovery by making sub-second RPO available in suitable environments. For our customers, that's another resilience option. For our partners, it's another reason to start a valuable conversation with their Oracle install base.
And after the conversations we had in Las Vegas, I think we'll have plenty more of those discussions over the coming months.
➡️ Learn more about Zero Data Loss for Oracle SE