AI Trends for 2025: Productivity, Integration, and Agentic AI

Machine-generated from the episode audio and not hand-corrected, so names and technical terms may be imperfect. The audio is authoritative.

All right, another episode of the CTO advisor podcast. It is just me. I'm Recording this the day that it publishes on December 18th 2024 It's been a wild couple of months. I have attended so many shows from Veeam to AWS reinvent and And everything in between IBM think The amount of data that I've consumed over the past few months has been Overwhelming is why I haven't consistently gotten a podcast out. I've just been traveling So we're going to do a podcast Recapping kind of what are the things that I think you should be?

considering based on my Experience with vendor shows CIO CTO Focus groups. What are the big themes going into? 2025 the biggest thing obviously is AI but not AI Like it was coming out of 2023 in 2023. We're talking about how big should Your cluster be to support emphasizing in your generative AI solution on the enterprise We've gotten a bit more clarity. I think and I think AWS reinvent is a great proxy for where customers are at and where the industry has moved to AI by itself is not a Application AI by itself is not a business solution the theme driving AI is Productivity gains right now.

So this ability to reduce cost in creating and consuming content this ability to Do chat box and summerization this ability basically to reduce costs around labor Because this is a massive opportunity is trillions of dollars Versus the three hundred billion dollar software industry. We're seeing a lot of attention from not just software providers and Software and IT DMS IT decision makers. We're seeing a lot of attention from The lines of business They want to reduce their cost. They want to increase their productivity.

They want to better their bottom line and the IT vendors are responding in Time at AWS reinvent. We saw AWS integrate AI into All the things and all the things integrated into AI We didn't hear much of any multi cloud conversation But that's not to say that multi cloud should be DS emphasized in 2025 How was AI? Heavily Represented in the AWS stack and not just AWS but other technology companies such as veil technologies It is much closer to talking about putting the rubber to road or meat on the bone from what's the value of the AI solutions that our people are building AWS case in point bedrock Amazon Q indexing services How do you build AI as part of your application strategy instead of just having the AI?

Strategy was the theme it is I called it seamless integration Nothing in IT is seamless But there is a well thought-out plan about how not just generative AI with Nova and Whatever other services such as anthropic etc is integrated in the AWS control plane But how do you? Build applications that can make calls to the services seamlessly turning the attention down the stack to a Dale who is now talking Data lakes and this ability to for storage to be more than just storage We saw this with AWS and s3 tables but AWS s3 tables was not the first one the market with this idea vast data with iceberg and Parquet file support this ability to enable AI from a low-level infrastructure service all the way up to high-level services such as Amazon Q for business and bedrock AI part of the process not just part of not just a standalone application We'll talk in a future about how companies are making this real and some of the real implications I've integrated AI into My workflow and I'm just a single person as we start to distribute this within our organization You'll see an acceleration of content.

You'll see higher quality content You'll see deeper insights with the same number or less people within our organization to do that. That is the driver for The biggest talking point of the year AI it has matured in 2024 even though for some people it doesn't feel that way multicloud So let's talk about keeping the lights on AWS rolled out their normal customers including well To my surprise Apple was on stage But not just Apple they rolled out capital one talking about how they're all in an AWS and they're taking a cloud-first Approach to everything they do But on the opposite end of the scale you have someone like Geico Geico has embraced a open stack.

I'm sorry open source slant I went to A Tech conference in Houston in which I engaged with some Geico representatives and they talked about how they're Replatforming every application within their environment to run either on open source or open source Services, so they're deploying open stack They're in Embracing Grafana for mark monitoring. They are rethinking all of their processes in the objective of being able to use modern abstractions and the in The service of being able to use AI and a highly regulated environment They've made the strategic decision in order to take advantage of AI and a highly regulated environment They have to rethink how they deliver IT When you rethink how you're how you deliver IT There's a in-between Geico capital one Multicloud we're not going to be able to build all of the services that a Geico is Building we're not going to have a Top down policy in which we can tell the business.

No, we have to build the application based on AWS services specifically capital one is taking a serverless first a lambda first approach if not lambda then Fargate if not fargate then easy to at the end and it allows them to build Operations that are catered around AWS. We don't have that luxury. Some of us have contracts with Microsoft Oracle We're dealing with our VMware landscape on-premises and then AWS is still very much the gorilla in the room for most of our organizations We've at putrim group started to do these CIO CTO Forums in which we bring in CIOs and CTOs from lines of businesses who can have one-on-one or one-to-many discussions with us and the vendor sponsor on their challenges and Multicloud continues to be a challenge I put out a Provocative blog post the other day in video saying that I don't believe in private cloud I don't believe in private cloud in the sense that you can take AWS put it on-prem Azure put it on-prem.

I'm not talking about outpost or Azure stack. Yes, you can take those things and put it on-prem We can't build what the cloud providers are building. We are not going to build a Open AI chat GTP alternative we can do cool stuff with quad and and granted and get close and have some of the capability but these Companies are innovating at a pace that we can't keep up with so our most innovative solutions Are probably best built on these platforms and when the enterprise Solutions the cot solutions catch up we can move those things from our Cloud providers who are cutting-edge on-prem, which we're seeing this repatriation But we have to deal with the resulting hybrid infrastructure in the hybrid operating model that we're in where we have VMware Hyper-V OpenShift KVM Bare-metal Kubernetes on-premises that's Interacting with services in the cloud lambda cloud cloud Actions or whatever the term for the serverless platforms to cloud providers have we're dealing with the security Implications of that.

How do we make sure that the lambda that's engaging with my Oracle DB or my sequel DB or Kubernetes on-premises have the Security context to be able to apply our security policies to these application flows How are we positioning our data in an equinix QTS digital realty? Etc to be able to take advantage of these cloud assets How do we catalog? Index all of this data so that we can make it available And some type of vector format so that we can leverage these AI services to get business insights last big theme of the year around AI is Agentic AI this idea that you're going to have a agent that acts on your behalf How is this different from the AI that we saw in 2023 early 2024?

Before we're using AI to again accelerate the consumption of data summarization of data the sorting of data the chat box of the world basic Data driven actions these Agents can act on Data so the difference is I can get a Agent that will book a flight based on my calendar not suggest flight options, but actually book a flight based on the count data within my calendar This is a huge opportunity for both enterprise software companies Cloud providers and infrastructure providers if we think about the nature of agents they are basically representatives of our organization If we have an agent act on our behalf, we're having the agent Do the work of the company?

That is very different than the file cabinet relationship that I have with my ERP provider that my ERP provider isn't doing work my ERP providers Providing me data and insights so that my people can make decisions and do the work What happens when the agent is embedded in my ERP system? So logistics has a request for additional material and a agent goes out and procures that Material without any human engagement or interaction What happens when you want that process move from your ERP system to another?

System or you want to replicate that in another business unit? Do you then need to go out and deploy that ERP solution to get that agentic AI capability? What happens when you need these agents to talk to one another do you go to a snowflake or a sales force or a Service now and have them built this capability and that capability is locked in or do you go to a Dell Technologies? Lenovo Cisco HPE And build your agentic AI With open source tools, which again move a little bit slower from our ability to consume this Capability, but now we own it now we can create this process that monitors our SAP ERP environment for these requests and then that AI Solution that we control we control the model we can Control the logic we control and place the platforms it can do this work on our behalf Huge debate and discussion that I've heard this year.

What's the answer to that the to that problem? I do not know speed to value having your enterprise application providers and cloud providers provide you this capability or Having the ultimate control of business processes in-house, but you need to hire for the skill set Not a tough. I'm sorry. Not a easy question to answer That's been my 2024 second half the we didn't get into the first half but that's been my 2024 in review and looking at 2025 and the things that I'm Keeping track of yes AI of huge importance, but because it's of use importance is Elevating the need for solid cloud based operations a solid data governance program and strategy and just fundamental primitives That's it for me for the year I am NOT going to do another podcast, but I will continue to create content through the end of year If you celebrate it Merry Christmas Happy Holidays Happy everything from the CTO advisor and future recruits