Bye 2024, Welcome 2025!
As the year draws to a close, I find myself reflecting on what an incredibly eventful time it has been—especially in the realms of AI and Microsoft 365 Copilot. The pace of evolution in these areas is faster than ever before, and witnessing how Copilot is reshaping the way we work has been nothing short of exhilarating.
This year, I’ve had the privilege of working closely with customers on Microsoft 365 Copilot adoption. From coaching and training sessions to innovating extensions and developing proofs of concept, it’s been a transformative journey. Helping organizations unlock the potential of AI and integrate it into their daily operations has been immensely rewarding.
Being actively involved with the Microsoft Tech Community—truly the best community in tech—has been one of the highlights of my year. I’ve had the opportunity to speak at events across Finland, Europe, and the United States, sharing knowledge and connecting with passionate professionals from all over the world. Meeting friends, both old and new, within this vibrant community has enriched my life, and I’m eagerly anticipating more of these meaningful interactions in 2025.Thank you everyone who made this possible!
Looking ahead, I know that 2025 will be a year filled with even more advancements in Copilot and AI. Customers are increasingly recognizing the value these technologies bring. However, they’ve also realized that a return on investment doesn’t happen automatically. True ROI materializes when a larger number of people understand how and when to use these tools effectively. This underscores the importance of creating relevant use cases that resonate with users and enhance their workflows.
I’m excited about the upcoming events that will kick off the new year. At the end of January, I’ll be heading to Tallinn, Estonia, for the Cloud Technology Townhall Tallinn (CTTT). It’s always inspiring to start the year by engaging with fellow tech enthusiasts and exploring the latest innovations.
In early May, I’ll be flying to the United States to speak at the Microsoft 365 Community Conference. Alongside my good friend Adam Deltinger, we’ll be hosting a workshop titled “Mastering Cross-Organization Collaboration with Microsoft Teams.” Additionally, I’ll be presenting a session on “Administrating and Governing Microsoft Loop.“ These topics are close to my heart, and I’m eager to delve into them with attendees. If you’re interested in joining us, you can register for the conference and save $150 with the code SAVE150.
As we approach the holiday season, it’s time to wind down and take a well-deserved break. The year has been filled with hard work, learning, and growth, and now it’s important to rest and recharge. I’m grateful for all the experiences and connections made this year and am enthusiastic about what lies ahead.
Wishing everyone a peaceful holiday break and a happy new year. Here’s to embracing the opportunities that 2025 will bring and continuing our journey together in the ever-evolving world of AI and technology.
Did that sound like written by AI? Probably yes and hopefully not too much. I deliberately used the Azure OpenAI Service o1-preview model to write this article, based on the information I gave to it, and did very little changes / additions. This reasoning model is just excellent! I have been trying it out on several occasions and I am truly impressed on what it can do.
After the first reply, I asked for a rewrite. Prompt is just a flow of my thoughts and not by any means a refined one.
GPT-4o is good, but o1 is way better with its 200K context capability. Answers I get from it are longer and better, it can understand much more complex inputs – I noticed if I just write prompts just like I would be speaking them.. it would give me great results. It is very easy to see how this will be changing the scene during 2025, as it will be more and more used. The model is currently in the preview but once it goes to GA it is interesting to see how it will be incorporated to Microsoft 365 Copilot. GPT-4o is faster (o1 is not slow either!) but there is a clear leap in the quality.
And it is stated in Azure AI Foundry models catalogue, there are very advanced situations where o1 will excel
The OpenAI o1 series models are specifically designed to tackle reasoning and problem-solving tasks with increased focus and capability. These models spend more time processing and understanding the user’s request, making them exceptionally strong in areas like science, coding, math and similar fields. For example, o1 can be used by healthcare researchers to annotate cell sequencing data, by physicists to generate complicated mathematical formulas needed for quantum optics, and by developers in all fields to build and execute multi-step workflows.
Using the o1 to write a blog article is just a small part of what it is capable of. How you can use it then? Just like any other model in Azure AI Foundry – deploy it and use it to power your custom agents! At the moment you need to apply for access, so you will need a good use case to get started with this.
Pricing is higher, as o1 will use much more computing power than other models. That is one reason why you should not always use the most powerful model, but instead choose the one meeting to your needs. Up to date pricing can be found at https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/#pricing
Comparing that to GPT-4o we can see o1 is roughly 6 times as expensive in the input and output.
When you have a good use case for o1,where results are important, then probably the price isn’t the biggest factor if it is saving people’s time big time – and/or delivering better and more correct answers compared to less expensive models. Perhaps you will have 1000 people using that several times a day, and you start calculating how high the cost will be. Remember to also estimate how much it will save time every day for those 1000 employees, and thus increasing productivity.
But the point of the pricing is: don’t just use o1 because it is there. Choose the model that meets your needs. For example there are o1-mini, gpt-4o, gpt-4o-mini and so on. There are , at the moment, 1820 models in Azure AI Foundry’s catalogue. If you need the reasoning and advanced model, o1 is certainly something you want to experiment with and try!
Published by
I work, blog and speak about Future Work : AI, Microsoft 365, Copilot, Microsoft Mesh, Metaverse, and other services & platforms in the cloud connecting digital and physical and people together.
I have about 30 years of experience in IT business on multiple industries, domains, and roles.
View all posts by Vesa Nopanen
