Introduction
Welcome to another episode of TJ's Technology Tuesday. Today I am covering the latest news from Microsoft. Why is this so relevant? For two reasons. First, most of you are already using Microsoft 365. Second, last week saw the Ignite conference. Ignite is Microsoft's developer conference – a bit like Apple's WWDC, it is where the most significant new developments are unveiled. And just like with Apple, there is an opening keynote that is reasonably accessible even for non-tech people, with just a few exceptions, followed by developer sessions that go very deep into the engine room. Since I work with business leaders who want to be digitally capable, I will give you – in my tried-and-tested, concise style – the things Microsoft presented that are relevant to you as a CEO, board member, or executive. I have taken some screenshots from the presentation. The setting was perfectly reasonable, and the content was highly interesting. Microsoft presented everything under the umbrella of three blocks with the overarching theme: "Democratizing Intelligence is Human Ambition – Co-Pilot, Agent, Ecosystem."
Which companies are leading in Artificial Intelligence?
Before they got started, they asked: "What are Frontier Firms? Companies that are truly out in front." Based on an IDG study, they said there are three types of companies. The first type watches and waits, saying: "This AI topic – we'll apply common sense and AI will pass, just like other trends before it." The second type tries AI here and there, experimenting a little in one place, dabbling a bit somewhere else. And then there are Frontier Firms, which are highly consistent: management decisions are made by humans, and all the doing – as far as possible – is handled by machines and agents. As a result, Frontier Firms are three times as profitable as those who merely observe. Against this backdrop, this is not just a nice-to-have; when implemented consistently, it is a significant lever on profitability. So what are the three areas Microsoft presented first? The first is having the right strategic mindset – what they call Human Ambition.
Exactly what I have been saying. The second is an improved Copilot, and the third is a complete agent ecosystem to go with it. When you look at the Copilot space, it always comes down to three things: first, the quality of the data on which it is trained and on which Copilot operates; second, how well it can make sense of that data; and third, how well it can remember things. On the data side, the OpenAI GPT model 5.1 has now increased its context window to one million tokens, and Microsoft has once again made its strategy very clear at Ignite: the goal is for Copilot to understand the entire business context even better. They call this Work IQ – a name I find quite apt, just as Copilot itself is a good term, since a copilot is not an autopilot. The whole thing is integrated not only into Microsoft applications but also into other data sources across the enterprise. And as a side note: those of you who have been frustrated when giving a command in the right-hand panel of Word or Excel – asking it to do something with the file – that is a capability that will be introduced soon and was among the new features announced.
Work IQ for Custom Agents
Of course, there is always a time lag between what is presented at keynotes and what actually arrives. The good news: looking at when Apple's first AI announcements were made compared to what is now slowly landing on the iPhone, Microsoft is actually moving considerably faster in terms of what is already available and what is on its way. The idea behind Work IQ is that I can work with all the data from the Office applications and beyond, and I can do all of this not just within a single Office application but across a wide range of functions that today are spread across different portals. The vision is a Work IQ for Custom Agents: the ability to work in a truly structured and effective way, with a Copilot available in chat, in applications, and in the areas of programming and administration. One highlight was a very impressive demo of using Word to fill in forms automatically based on information from other sources. That is a use case I find extremely practical and it will be available soon.
News on the Topic of Security
Not glamorous, but highly productive. They also presented updates in the area of security – including Microsoft Defender, shown in a screenshot – where AI is used to detect phishing emails, combined with decision trees, which I consider a very sensible application of AI. The result is a concrete recommendation: is this critical or not? Another improvement on the horizon – and perhaps you have already been frustrated by this – is a feature that already lets you summarise emails via the Microsoft Copilot for Work and even have drafts written for you. The next step, which is coming soon, is the ability to send directly from there as well. Features that are currently offered by third-party tools will be increasingly integrated, so that you can control your inbox and calendar entirely by voice. Even the agents in Copilot Studio, even when added as a tool, still struggle a little with this. And then the Security Copilot – very impressive, giving you suggestions and making things visible even if you are not a deep technical expert.
With Microsoft's "Lovable" You Write Applications Using Prompts
Here again I want to highlight one example from a medical practice. The logic is that you can have voice conversation notes – similar to what you might know from Plant – integrated from a system. You can already connect voice as a trigger in Copilot Studio today, and they showed how this can be embedded even more intelligently in applications. On the topic of application development and programming: Lovable – which I think I have mentioned here before – is a Swedish platform for building applications without traditional coding. Microsoft is now introducing something similar, letting users write applications with prompts. They showed a preview example; and "Frontier" is essentially what in my own tenant often appears as a Microsoft "Lovable" – these preview features. From there, you can build an application with a simple prompt. These things still look a bit Microsoft-style by nature, whereas Lovable itself tends to produce something a bit more visually polished. But I will also test the Microsoft version once it is available to me – to see whether I can say, as I can with Lovable: "Take the Apple iPhone page as a reference and adapt this UI, this user interface, for my application."
Further Alternative AI Models (Claude & Co.)
I assume that will be possible. So here truly more examples, and the Copilot Agent can also be extended here. One extension I find particularly impressive: the ability to use not only Microsoft models and OpenAI models, but also other models. Claude was explicitly brought on stage – the company Anthropic, headquartered in Paris. In other words, a French, European model that you can integrate. The idea Microsoft is pursuing – and I find this very intelligent – is to position themselves as the provider of the work tools and work environment: secure, protected, yet open to additional AI models. This is also true for programming: for those working with GitHub who want to go a little deeper, there were sessions showing that the Copilot integration in GitHub is even better now. Interestingly, I have always said in the past: "GitHub is great, but also look at Claude from Anthropic in parallel, because for coders that has always been the benchmark."
And what has Microsoft now done? They have partnered with Claude. That means you can now use Claude within GitHub. The best-of-breed world is coming together. Explicitly: Anthropic – the coders' dream model, as I always say – has been integrated here. In the Microsoft AI Foundry – the foundation I will say a few more words about in a moment – you can choose from various models, including Deepseek, a Chinese model, but always within the data-protection boundaries that you can configure. There is also the option of an AI Model Router: the system itself decides, to some extent, which model is best suited for your particular request. I find that fantastic, because with the growing number of models, especially when I test local AI models, it is extremely difficult to decide which model is best for which task. Today I often use Claude from Anthropic to answer exactly that question: given this task, which locally installed AI model should I use?
And that will surely be one of the components Microsoft has integrated into its AI Model Router. They also presented their strategy in the form of an Intelligence Layer, explaining that there is context, there are decisions, and there are the corresponding actions behind them. Very nicely put, they said that complementing Work IQ there is the Fabric Copilot – which I will explain in a moment with a diagram – and Foundry as the foundation where the various AI models reside. They also showed the ability – which was quite rudimentary at the start – to have a proper agent administration console, where you can see which agents are deployed, which may even have security risks, and how they are being used. From there you can approve agents, see where problems exist, and fix them. And the great thing is that when I look at a multi-step workflow, not every agent is simply created freely; instead there is a review process, an approval process, before agents are released.
Microsoft Fabric is a Master in Data Engineering, Analytics, Science and AI
You can also see how agents are interconnected, where the links between agents are, and which ones are used most frequently. And then Microsoft Fabric – here too a screenshot from the publicly available YouTube video. The idea behind Fabric is that it brings together all the power components of the data world: Power BI, Real-Time Intelligence, databases. The concept is a Data Lake approach. There has always been this distinction between Data Lake and Data Warehouse: Data Warehouse holds structured data, Data Lake holds unstructured data. Microsoft is increasingly bringing these together and saying: if you also have data sources outside the Microsoft world, the idea is to consolidate everything so that I have all the tables from various different systems and can run cross-query requests across all of them. This gets quite technical, but I hope you get a sense that Microsoft has, in my view, a very strong strategy: not only expanding Copilot's capabilities, but also looking at what competitors are doing intelligently.
What other data sources are companies working with? They are not working with Microsoft alone – connectors have existed for some time. But the idea – and this is also shown clearly at the end, with phases like Enrich, Reinvent, Reshape, Bend the Curve – is that connectors are one thing, but when you intelligently connect them all, you arrive at what we discussed earlier: Work IQ. And all of this very intelligently designed. I would like to close the loop here by saying: companies that understand AI not merely as a tool to experiment with, but that think strategically about how to consolidate their data on a system that can also integrate different platforms, how to set up a security concept, how to build agents and applications integrated into Office applications and beyond – these companies are the ones that will become Frontier Firms. Firms where humans make the decisions, but machines do the doing. On that note, I wish you as a decision-maker a successful week.
Conclusion
I hope I was able to give you a little more background on all of this. You know I am someone who can also programme, but I always say: even as a business leader, you need a certain level of understanding. You do not need to know how to write SQL queries, but you do need to be able to make decisions – for or against Microsoft 365, for example. I am a Microsoft partner, but even if I were not, I would give exactly the same recommendation: Microsoft 365 is, in my view, currently the most sensible platform for enterprise use. And when I look at the steps they are taking towards AI – large, confident steps – and what I see arriving in practice, there is still today perhaps 90% of situations where I say Microsoft is really, really good. And for the 10% where you sometimes hit a wall, I am happy to support you with consulting, implementation guidance, and training. Tomorrow, for instance, I will be at a car dealership, looking at how manual processes can be streamlined using Power Automate and Microsoft 365's AI capabilities – so that more business can be done with the same team.
That is when technology makes sense. Your Personal IT Coach for Business Leaders, Thorsten Jekel.
Key Takeaways
- Microsoft Ignite is the annual developer conference where major updates for Microsoft 365, Copilot and AI are announced.
- According to an IDG study, so-called Frontier Firms are three times as profitable as companies that merely observe or sporadically experiment with AI.
- Frontier Firms leave decisions to humans while handing all operational execution, as far as possible, to AI agents.
- Microsoft is enhancing Copilot with the concept of "Work IQ": Copilot is designed to understand the entire business context and also connect data sources outside of Microsoft.
- The OpenAI model GPT-4o 5.1 now features a context window of one million tokens, significantly improving the quality of Copilot responses.
- Microsoft is integrating alternative AI models such as Claude (Anthropic) and Deepseek into its Azure AI Foundry – including data-protection boundaries and an automatic Model Router.
- Claude from Anthropic is being integrated directly into GitHub, so developers can use the leading coding model within the Microsoft ecosystem.
- With "Lovable by Microsoft," users will be able to create applications using prompts, without traditional programming.
- Microsoft Defender uses AI combined with decision trees to detect phishing emails and provide concrete action recommendations.
- Microsoft Fabric connects Data Lake and Data Warehouse and enables cross-query requests across various internal and external data sources.
Frequently Asked Questions
What is Microsoft Ignite and why is it relevant for business leaders?
Microsoft Ignite is Microsoft's annual developer conference, comparable to Apple's WWDC. It is relevant for business leaders because it is where the most significant new features for Microsoft 365 and AI are unveiled – developments that directly affect day-to-day business operations.
What are Frontier Firms and how much more profitable are they?
Frontier Firms are companies that deploy AI consistently: humans make the decisions, while machines and agents handle all operational execution as far as possible. According to an IDG study presented at Ignite, these companies are three times as profitable as those who merely observe AI developments.
What is Work IQ in Microsoft Copilot?
Work IQ is a new concept in Microsoft Copilot that enables it to understand the entire business context. It connects not only data from Microsoft applications but also additional enterprise data sources, enabling smarter and more context-aware responses.
Which AI models can be used within the Microsoft environment?
In Microsoft Azure AI Foundry, in addition to Microsoft's own and OpenAI models, alternative models such as Claude from Anthropic and Deepseek can be used. All models operate within configurable data-protection boundaries.
What did Microsoft announce about Claude from Anthropic at Ignite?
Microsoft announced a partnership with Anthropic at Ignite and integrated Claude into GitHub. This allows developers to use the AI model widely regarded as the benchmark for coding directly within the Microsoft development environment.
What is Microsoft's AI Model Router and what is it for?
The AI Model Router is a feature in Microsoft Azure AI Foundry that automatically decides which AI model is best suited for a given request. This significantly simplifies the work, as users no longer need to select the right model themselves from the growing number of available options.
What is "Lovable by Microsoft" and what can you do with it?
Lovable by Microsoft is a new feature that allows users to create applications using prompts, without any traditional programming. It takes inspiration from the Swedish platform Lovable and is being gradually integrated into the Microsoft environment.
How does Microsoft Defender use AI to improve security?
Microsoft Defender combines AI with decision trees to automatically detect and classify phishing emails. The system then delivers concrete action recommendations, indicating whether a detected threat is critical or non-critical.
What is Microsoft Fabric and how does it help with data analysis?
Microsoft Fabric brings Power BI, Real-Time Intelligence and databases together under one roof, connecting Data Lake with Data Warehouse. It enables cross-query requests across various internal and external data sources, including those outside the Microsoft world.
Is Microsoft 365 worth it for companies given the AI developments?
Thorsten Jekel recommends Microsoft 365 as the most sensible platform for enterprise use today. The consistent AI integration – from Copilot to agents and alternative AI models – makes Microsoft 365 a powerful lever for increased productivity and profitability.
Tools & Resources Mentioned
- Microsoft 365 – the recommended platform for AI-powered enterprise use
- Microsoft Copilot (AI) – Microsoft's AI assistant with Work IQ functionality for business context
- GitHub – Microsoft's development platform, now with Claude integration
- Anthropic / Claude – AI model integrated into GitHub and Azure AI Foundry at Ignite
- Lovable – Swedish platform for building applications with prompts, which inspired Microsoft's new feature
