A Jump in the June Pool of Data

Well hello there again! It's that time of the month when I uncover the pool and invite you to jump in to swim in some data.

This month there's more AI, of course, but I've also got some data data (yeah, isn't that grammatically fun?) and a few tidbits on modernization. Of course there's our own State of AI Application Strategy, which we just released this month, that's chock full of interesting findings. Overall the theme fits a summer swim: there's the folks that just cannonball right in and then there's those that take a more deliberate approach and slowly but surely wade in to acclimate to the water temperature.

And apparently there's more of the latter than the former based on some research from Couchbase that found nearly two-thirds (64%) of IT leaders believe that other companies have rushed to adopt the technology without properly understanding how to use AI safely.?

What we're approaching for generative AI is the same "big hill" every emerging technology faces after the initial race to grab all the low-hanging fruit. Organizations everywhere are recognizing the need for a strategic plan that includes modernizing their infrastructure to support the unique needs of this new, amazing technology.

An AI Application Architecture is Emerging

Over the past few months the "AI Application Architecture Pattern" has become more clear. Part of that architecture includes reliance on RAG (Resource Augmented Generation) to deliver greater value and efficacy for applications that make use of foundational (or stock, if you prefer) LLMs. That's why it was no surprise that DataBricks found vector database adoption has grown 377% year over year, especially as 70% of their respondents leveraging generative AI are using tools and vector databases.

But part of that pattern includes the expansion of the modern application archetype to include a second "tier" where inferencing servers are deployed and scaled to serve the chatbots and content generation and search capabilities companies are already deploying (so says our research). That's not a bad idea, because 60% of IT decision makers worry about ensuring their organization has sufficient compute power and datacenter infrastructure to support GenAI according to Couchbase, and separating the tiers gives organizations a way to build out (and modernize) the compute and infrastructure they need without significantly disrupting their existing applications and supporting infrastructure.

The good news is that organizations are putting their money where their modernization efforts are. RedHat found that 59% of budgets are in service to modernizing existing infrastructure and/or applications. And how are they doing that? Well, RedHat tells us that AI is being used to modernize applications. Over 75% of organizations they surveyed are doing just that, with 42% adding AI to existing legacy applications to modernize them.

In fact, we're seeing a lot more use of AI to modernize everything from infrastructure to security to business functions than we are a focus on delivering AI apps.

Putting AI to Work Within

The use of AI to modernize and optimize infrastructure and operations is not a new concept. Our own research has shown that using AI for IT operations (some call this AIOps) has been a thing for years now. I don't need external validation, but SAP provided some in case you do. Its research found that the top three uses of AI were:

o?? 69% IT infrastructure management

o?? 52% cybersecurity alert prioritization

o?? 49% marketing and sales ops

That makes sense; AI is all about efficiency and productivity. That's not just me, Simform said it to in their survey, too, which found the primary metric for measuring ROI is ‘Efficiency and Productivity Gains’ for 56.1% of the participants.

Yeah. AI and Automation go together like, well, two words that start with A. But seriously, the use of AI is going to have the biggest impact on automation since Henry Ford fired up the first assembly line.

The thing is that generative AI alone won't do it. It's going to take predictive AI, as well, because before you can automate something you need a trigger, and the trigger in operations will be some kind of insight derived from lightning-fast analysis of real-time data. That's the predictive AI side. But predictive AI can't act, that's where generative AI comes in. Together, these two types of AI are going to change the very fabric (pun intended) of enterprise infrastructure, no matter where it may reside (core or cloud).

The AI Hype Train is Chugging Along

The AI hype train is not slowing down, but it's not speeding up anymore, either. That's a good sign, because we've been bombarded with AI this and AI that for the past year and quite honestly, at times it's overwhelming. The FOMO being thrown at the market is strong at this point, but I'm seeing signs that a lot of organizations are ducking that FOMO and approaching AI as the serious transformation it will be.

Slow but steady wins the race, and building houses on sand is never a good idea. So kudos to everyone who is thinking strategically, especially in the summer when the pool is open. I mean, we need to have our pool time, right? RIGHT?

Of course we do. Now get out there and touch some water!

Stay cool until next time.