Josh:
Welcome to episode 59 of The Leading IT Podcast. This is a huge one! Today, we’re going to break down the game-changing announcements from OpenAI and Microsoft and how they relate to IT leaders in business. This podcast is for Australian IT leaders looking to stay up to date with the latest news and trends in AI, cybersecurity, cloud, and infrastructure strategy and leadership. Your hosts are Tom Leyden, the CIO at Longview, and me, Josh Rubens, CEO at consulting firm Empyrean IT. We tackle the fast-changing IT landscape from both sides of the client-vendor paradigm with pragmatic and actionable advice. Good day, Tom.
Tom:
Good day, Josh. Uh, you’re right about fast-changing. There’s a lot to talk about.
Josh:
Yeah, it’s massive. I feel like the world changed last week for me. The more I looked at it, the more I thought, “Oh goodness me,” and a few other choice words that I won’t use on a PG-rated podcast.
Tom:
Amazing, isn’t it? I guess what we get to do is try to help people navigate through these changes and talk about the pragmatic implications of it as well.
Josh:
Yes, exactly. I don’t think I’m overstating it. I think as we go through it, we’ll explain why I think some of these changes are going to have such a big impact. Some of it is obvious, and some of it is less so. We’re not going to do our usual news because there’s so much to get through. If we can’t cover everything today, we might have to continue next time.
So, the three big things were the OpenAI GPT-4 Omni release, the new model that came out last week, the Copilot Plus PC announcement from Microsoft, and the Microsoft Build 24 conference that happened. Microsoft was a bit sneaky—they had the Copilot Plus PC announcement one day, and literally the next day was the Build conference. There was a lot happening, and I think the industry and the market, in general, are believing in the story. I mean, if you look at the NASDAQ and those particular stocks, they’ve gone through the roof. Nvidia, in particular, hit a three trillion-plus valuation this week.
Tom:
Yes, yes.
Josh:
Hopefully, you bought in, Tom, when I said you should.
Tom:
Um, yeah, maybe. But yeah, there’s a list of things I should have bought, Josh. There’s a huge list, and I think Nvidia is right at the top of that list, actually.
Josh:
Okay, I did get in. But no, it’s good. Who knows what’s going to happen, but I think the market is definitely believing what Microsoft is selling, and it’s pretty huge. Let’s get into the GPT-4 Omni release. Tom, did you watch the video?
Tom:
I did. You know, the fanfare and whatnot. Microsoft does these releases with a lot of excitement. Everyone believes it’s the Steve Jobs of news. Are you a believer, Josh? Do you believe?
Josh:
I am, I am. What are the highlights for you, mate? Do you want to run through your highlights?
Tom:
Yeah, so the GPT-4 Omni from OpenAI is different in that you can interact with it using voice, video, and chat in one model. Previously, these were siloed into separate models, which meant longer response times and higher costs. Now, they’ve merged these capabilities into one model called the Omni model, and they’ve also released a desktop app. Now you can interact with it within the flow of your work. It’s a little bit like Siri or Alexa, but on a huge amount of steroids. You can have voice interaction with the model. The audio input response time is now 320 milliseconds, which is on par with a human response time. You can have a genuine chat with the model. You can change the voice, and it feels like you’re talking to a person, which is amazing.
Josh:
The example I saw had language translation as well. You could change language halfway through your conversation, and it just spoke Spanish to you. That was super cool, right? The integration of those models is really good, but it’s now cheaper and faster to run, which is important.
Tom:
Yeah, it’s basically half the price of GPT-4 Turbo, so it’s getting faster and cheaper. The voice interaction is one part, and then there’s the video interaction. These are real game-changing things. In the video, you can show things to ChatGPT using the camera on your computer or mobile device. For example, you can have ChatGPT as a virtual person on a call, show it a spreadsheet, and ask it questions about what’s on the spreadsheet. It can analyze what it’s looking at and talk to you. It’s almost like having an expert because it can look at what you’re showing it on the screen.
Josh:
It’s mind-blowing. You can also show it things on your camera. One demo was a guy on a website for mountain hiking gear. He asked if the shoes he was considering were good enough for climbing a mountain, and it said no and recommended another pair, which it then added to his shopping cart. It’s incredible.
Tom:
Yeah, the commercial aspect was like it’ll help your customers buy more stuff. But I think there are broader applications. Think about people dealing with frontline responder issues. They’ve got cameras, and the AI can help identify issues with a particular problem, medical emergency, or any situation in an environment. That’s when it gets really interesting.
Josh:
So, instead of the interaction with AI being through a prompt and chat, it’s now through voice and video. Satya Nadella said the world is the prompt, and that’s the game-changer. GPT-5 is coming out in three months and will be even faster. The pace of change here is amazing. You don’t have to think long and hard about all the business and commercial use cases for voice and video interaction with AI. It’s smarter than humans and can analyze what’s going on around you. It’s incredible. You can interrupt the model, it can recognize emotion, and it can generate different emotions in how it talks. You can interact with it through video, camera, or screen. For example, a policeman with a body camera could get live information from the camera about what’s going on around him.
Tom:
If you think of a future where sometimes you call up and don’t know if you’re talking to a human or a call center, and the call center sounds human, has all the information, and knows exactly what they’re talking about. They’ve already verified you because they understand your security questions. This is a complete game-changer for interacting with your customers. All those human-to-customer touchpoints need to be rethought. How can you start exploiting this technology to make your business front end more efficient?
Josh:
Yeah, it can even pick up on whether you’re having a good or bad day. You can train it to have humor or no humor. There are many parameters you can put in. Over the next few months, we’ll see a lot of things come out around that. This led to the Copilot Plus PC announcement. You might think it’s just about a neural processing unit on a PC, but it’s much bigger. Did you look at it, Tom?
Tom:
I did. Maybe just take a step back and explain what an AI PC is because many people might think it’s just a new key on a keyboard, but it’s much more.
Josh:
Yes, it’s a new category of PCs with new hardware, including a neural processing unit. The first one to market is the Qualcomm Snapdragon ARM processor, but it also has a normal CPU and a GPU, making it a lot faster and able to process AI jobs locally. They’ve rewritten Windows 11 to support small language models running locally on your device. Instead of going to the cloud for AI processing, many tasks can now be done locally. There’s a new Windows runtime and new applications. This is a pretty massive thing. For end users, Microsoft says this is a genuine reason to buy a new PC because of the new capabilities.
Tom:
Yeah, maybe let’s talk about that.
Josh:
Yeah, go on.
Tom:
It’s a fascinating architecture. The NPU takes a bunch of predictable AI workloads and runs them more efficiently and longer on a lower battery, which is better for laptops. But the interesting thing for PC fleet buyers is when to start buying these. NPU processing is good now, but what’s coming down the pipeline is significantly better. I’d suggest giving this 12 months for the software to catch up and the processing power in the NPUs to accelerate. In 12 to 18 months, it’ll be something to really consider.
Josh:
Yeah, I agree. Not going first is a good idea. Intel and AMD are coming on board, as are all the normal hardware vendors like HP, Dell, and Lenovo. It’s capable of 40-plus trillion operations a second, which is fast, and 58% faster than the new MacBook Air M3. Some key applications leverage a local Windows semantic index, building a profile of what you do across all applications and browsing. It’s like a snapshot, but it takes you back into the application. Security is a big focus, and they say it’s only on your device, and you can delete snapshots. That’s interesting.
Tom:
They’ve got to demonstrate the upside of recall. There have been some public security issues with recall, and the fear factor resonates with many people. I think they’ve got to show why this is a good thing for you.
Josh:
Yeah, finding stuff quickly is a pain, and this solves that problem. Adobe apps are supported, and there’s an emulator called Prism for app compatibility. Live captions can translate 40 languages in real-time during a meeting. That’s pretty cool.
Tom:
Unreal. Who would have thought you could have a UN-style meeting with real-time translation?
Josh:
The video showed four people each speaking a different language, all translated into English in real-time. There’s also co-creator, where you can sketch on your device, and ChatGPT creates an image for you. CapCut allows you to change the background of a video. Windows Studio Effects can change background contrast and more. It’s changing what a computer can do for you.
Tom:
The real-time processing is interesting. For graphic design types, they’ll be knocking on your door demanding these things.
Josh:
Yeah, for sure.
Tom:
Also, for executives or senior managers doing large-scale meetings or dealing with multinational businesses, the language translation will dramatically improve conversations.
Josh:
There’s also a virtual assistant for other applications, and to start using the 4 Omni stuff, you’ll need these NPU devices. Think about the voice and video prompts you can have.
We don’t need to get into the ethical considerations now, but it could definitely impact many people’s jobs.
Tom:
We don’t know yet. It’s a discussion for another day, but the technology is here, and you should be riding the wave.
Josh:
If you’re a CIO, it’s your job to look at it. These devices are coming out on June 18th at a thousand US dollars, which is pretty good. 22 hours of battery life, Tom. That’s pretty good.
Tom:
Do you believe it?
Josh:
Well, I could say they showed videos, but yes, I believe it. That’s a good question. Other things include support on Intune and Autopilot. It’s ready to go with built-in security, advanced security with the Pluton security processor, and Windows Hello support.
Tom:
They’ve got to be prepared to defend the security aspects, especially around recall. Some people might be concerned about taking a shot of everything they do every five seconds.
Josh:
Yeah, but the final version has advanced security, and it’s ready to go. They’re shipping the Pi3 state-of-the-art small language model with 3.3 billion data points.
Tom:
Small language model, but that’s pretty good for a narrow use case.
Josh:
Exactly. Moving forward, in an enterprise, you have many options. Azure or AWS AI tools for highly customized applications, OpenAI for consumer local devices, M365 Copilot for your 365 stuff, other copilots, and small language models for different use cases. You might even have a private cloud for certain organizations. It’s not either/or; you use the right thing for the right use case. What are your thoughts on that?
Tom:
It’s about bringing your business to AI. Develop your own models around your unique information sets and capabilities. Integrate those with the broader AI ecosystem to create a new way of doing business. It always goes back to knowing where your data is, the quality of your data, and the quality of your tech team. Make sure they understand machine learning, AI, programming, and security. They need to think about how to turn this stuff into the AI-led business of the future.
Josh:
AI is going to be infused into everything. We need to reimagine our business processes to take advantage of it. Should we change tack and move into the enterprise and talk about the Build announcements?
Tom:
There’s a lot to get through, so maybe a high-level view of Build without too much detail. There was heaps to digest.
Josh:
Yes, a lot. I want to cover off what I think are the main half-dozen things that are critical for IT leaders to think about. First, Microsoft announced the availability of GPT-4 Omni the day after OpenAI did. Just to show how closely aligned OpenAI and Microsoft are. You can go up to Azure AI Studio, select it, and get going today.
There was a lot, but the main area for me was the concepts of the Microsoft Copilot Studio and the Microsoft AI Studio. Let’s differentiate between the two. The Microsoft AI Studio is utilizing the models in Azure, where you can choose models, orchestrate them with your data, and build them together. It’s more of a full-code developer platform for high personalization and integration. The Microsoft Copilot Studio is more of a low-code platform for leveraging things within M365 or Power Platform, where much of the stitching together, like security management, is already built for you. So, you don’t need to be a hardcore developer. Does that make sense?
Tom:
Yeah, it does. Some parts you unleash on the business for people to do low-code stuff, and some parts empower your IT org.
Josh:
The two things that stood out to me were the concept of creating custom copilots that can act as an agent within your business to deliver end-to-end asynchronous business tasks based on triggers. This changes the strategy of AI with Power Platform from building bots to having autonomous agents that deliver end-to-end business processes.
They’ve launched templates for IT help desk, employee onboarding, and more. These autonomous bots can guide users through processes, integrate with HR, and set up meetings. It’s like having an expert in different areas baked into your processes. The impact on business efficiency and cost reduction is significant.
Tom:
Most people might think, “That sounds great, but it wouldn’t work here.” They need to see it in action to understand how smart these things can be.
Josh:
It’s like a power app but run by a bot with AI. It knows what to do next. You ground it with your information, connect it with APIs, and it runs end-to-end business processes. It’s not a huge change from building power apps but has a massive impact on business efficiency.
Tom:
Yes, it’s exciting. This is the fun part to exploit. Get a team member in your team that handles compliance, project planning, or meeting minutes. It reminds you of what was discussed last week and tracks action items. Then explore having bots specialize in specific aspects of your work.
Josh:
The vision is to have bots that talk to each other and become team members. You could have bots named for different roles like finance or HR. It’s not a huge leap from building power apps but has enormous potential.
For example, a bot can get an email from a customer, look up their chat history, respond, check stock levels, and process the order. Happy days.
Tom:
Exactly. Maintenance requests, IT help desk—it’s all there.
Josh:
Also, employees can create their own copilot in SharePoint to interact with their data using natural language and extend it to third-party data. You can share that copilot with team members in Teams. That’s going to be big because finding stuff in SharePoint has been a pain.
Tom:
100%. Procedure manuals and how we do stuff here—it’s all about simplifying that process. Instead of going to the SharePoint website, just talk to the bot.
Josh:
Exactly. Now, let’s talk about the data. Did you see the real-time intelligence in Microsoft Fabric announcement?
Tom:
I did. Really interesting. It’s what we’ve been waiting for—to take action quickly when you see events occur. Super exciting.
Josh:
To cover the main points, Fabric is a SaaS solution that aggregates all Azure data services. There’s a real-time hub to ingest data and events, connectors to different cloud sources, real-time dashboards, and AI-driven data activator to respond to changes in data with business processes. It’s called the event-driven fabric, responding to events without needing eyes on the screen.
Tom:
Exactly. Stock levels, inbound call notifications, website visits—there are many applications. IoT space is enormous.
Josh:
There’s a copilot in Fabric for conversational language to create data flows and pipelines, generate code, and interpret data. It’s another huge announcement.
Tom:
Microsoft is trying to be your best friend. Give them everything, and they’ll help you out.
Josh:
They’ve got connectors to Databricks, Snowflake, and other platforms.
Tom:
Exactly. Now, in Azure, there’s a copilot for application troubleshooting, fixing errors, reducing costs, and SQL queries. It helps optimize your Azure environment.
Josh:
Great. It lifts technical expertise to the next level.
Tom:
Absolutely. Power Platform also got a big security update with a security hub, posture assessment, deep visibility tools, privilege identity management, native GitHub integration, canvas app co-authoring, and copilots on mobile. Lots happening there.
Josh:
Yes, lots of updates. I think we’ve covered the main points. Anything else, Tom?
Tom:
Just the net Aspire piece for software development, making it easier to deliver code.
Josh:
Yes, lots there. I encourage people to seriously look at the Copilot Plus PCs, get into GPT-4 Omni, and start thinking about how to integrate them into your Power Platform. Do proof of concepts with your data and applications to stay ahead.
Tom:
Think big and think small. Start imagining your future business and demonstrate small wins to show what this can do. You need to be there quickly.
Josh:
If you’re not doing it, your users will. This is prime time.
Tom:
How many times have we said that?
Josh:
This is prime time. You’ve got to do it. Get your data and security in order, but start now. Do some fun things. It’s an exciting time.
Tom:
It’s a good time to be in this seat. You are in the lead.
Josh:
Yes, exciting times. Do it before your users do, and stay relevant.
Tom:
You’ll be CEO if you pull this off.
Josh:
Exactly. Thanks, Tom. That was a good conversation. Was I overstating how significant the changes are?
Tom:
No, it’s huge. We should talk about what other companies are doing in upcoming podcasts.
Josh:
All right, I’ll go have a look at Google. We can discuss what others are doing.
Tom:
Good to see you up and about, Josh. I like it.
Josh:
Thanks, Tom. Good to see you too. Thanks, everyone, for listening. Please give us a five-star review so we can spread the word and help you stay up to date. Have a good long weekend. Thanks, Tom.
Tom:
Thanks, Josh. Speak soon.
