In short
Our recap of the European Collaboration Summit in Cologne. Discover insights on building & securing M365 Copilot Agents, data governance, and the future of IT.
On May 6th and 7th we visited the European Collaboration Summit in Cologne, Germany.
As we went back and forth between Expo hall, conference hall and conference rooms, inside the vast Koelnmesse venue, one thing was clear: A.I. and Agents are more than ever present in our daily lives, and affect us especially as IT-professionals.
A.I. everywhere
What are the possibilities of Microsoft 365 Copilot Agents and how do we build and secure them? How to use Purview as a tool to achieve the desired security levels?
How do we govern our data in SharePoint?
What is our role in the future, with an ever-advancing digital co-worker at our side?
Will it simplify and ease our work, or is it here to eventually replace us?
All these questions we have when thinking about A.I. were handled during several sessions at this event.
A.I. to shift the way we work
A.I. and Agents have evolved at a surreal pace.
Whether you want to use them for intuitive interaction between users and data, or to help you on a daily basis as an IT professional.
One of its uses might be to automate user onboarding or offboarding, as Julian Rasmussen explains during his session “Entra ID Management Made Easy: Learn How to Automate Access, Onboarding, Offboarding, and Movers”.
A combination of SharePoint Lists, Forms and Power automate provides us with a new user to create, Entra ID Lifecycle workflows help us to manage onboarding steps, and Access Package is used to give, approve, review or remove access.
Data exposure and security measures
An Agent might even appear as your co-worker.
Before there were a few types of identities, we were all familiar with: Users, Apps and Devices. Now there is a new form of identity: the Agent.
As an Agent can act on a user’s behalf, coordinate actions with its own authorization, or even provision its own access and resources, new tooling like Microsoft Entra Agent ID are brought into life, as the need for secure A.I. Agent access rises.
According to Bram de Jager during his “Securing M365 Copilot Agents; Governing the New Frontier”, our sensitive data is more vulnerable than it used to be. Until recent, a certain level of Copilot prompting knowledge was necessary, now the threshold has been lowered because the agent responds conversationally. The real risk is in fact not the A.I. or the agent, but the access model of your data and the way users interact with it.
Key takeaways: Start with assessing how easily data is discoverable in your organisation, use sensitivity labels. Secondly protect your data, use Purview’s Data Loss Prevention. Also protect interactions that can be made with your data, act on your sensitivity labelling. And lastly govern your data.
In other sessions the importance of data governance was acknowledged, for example during “Best practices deploying Purview” by Oliver Sahlmann and Dominik Höfling.
Adaptive Protection, a combination of Information Protection, Insider Risk Management and Data Loss Prevention, could save you from a gnarly newspaper headline, explaining how you fell victim to involuntary data exposure.
Ease your employees into adapt a secure mindset, inform them, plan ahead, do try-outs with small user groups in simulation mode, collect feedback and finally rollout and enforce your security solution company wide.
Be aware of chronic oversharing, Antonio Maio explains during “Guardians of M365: Tips for Effective SharePoint Management and Data Governance!”.
Use SharePoint’s Restricted Content Discovery to shield sensitive data from being discovered by Copilot.
Try to watch out for easing into a false sense of security when your data is protected with Everyone Except External Users. Use SharePoint Advanced Management capabilities to generate Data Access Governance Reports for SharePoint sites, to detect Permissions & Sensitivity.
For larger enterprises, consider using Microsoft Graph Data Connect for SharePoint instead, where the focus leans towards Oversharing & Permissions.
And lastly, beware that even though you applied security measures and government, those measures might protect sensitive data against being discovered by Copilot, but not necessarily from other A.I. tools, used inside your organisation.
This is a problem that can be helped by using the correct sensitivity labels in combination with permission levels.
The future of IT companies and IT professionals
So, building Agents, securing them and possibly have them as a form of co-worker with its own identity. How will this affect us as IT professionals and the companies we work for?
Chris O’Brien tries to answer this question in his session “From boardroom to build room on Microsoft. Staying valuable in the A.I. shift”.
We have all seen the news.
The Guardian, The Japan Times and others headline: “AI poses unprecedented threats Congress must act now (Bernie Sanders)”, “What happens when paid work disappears?”, “With A.I. Set to Destroy Millions of Indian Jobs, Traditional Career Advice is a Death Trap”, and more.
Today, Tech companies heavily rely on billable hours.
But according to O’Brien, the use of billable hours as we know it, is going to change over the next five to ten years, and only stronger companies will survive.
This assumption is based on a precedent: the digital evolution of the internet over the last 30 years, resulted in a few dominant companies that emerged.
For example, free use of search engines, e-mail and social media, all largely funded by ads, and smartphones have a payable app for everything.
Companies like Google, Meta, Microsoft, Apple, Android and Amazon dominate the market. Those companies have something in common: they had early mover advantage, network effect, and distribution.
No matter how much competitors seem to invest, they still seem to lose to these dominant players, because a handful of platforms can decide what is discoverable and what is surfaced.
The journey of A.I. will have a similar effect, but instead of 30, it only took a mere 3 years to cover the same progression.
Whereas 5 years ago, A.I. had trouble spotting the difference between a chihuahua dog’s head, and a blueberry muffin; today’s story is something quite else. And those new capabilities also have a significant effect on the stock market.
Some recent events that resulted in large stock drops: in July 2025 Microsoft lays of 20% of its Legal team to save costs and invest in A.I.
In January 2026, Anthropic launches Claude Cowork. On February 3rd 2026, Anthropic releases 11 industry plugins for legal, sales, finance, marketing, customer support, and others. On the same day, Open AI launches its Frontier Agent. In short, $285 billion in stock dropped in 48 hours.
For more details visit https://deathbyclawd.com
To survive in the future, organizations more likely to survive are those who have domain expertise, a distribution or network effect, a strong customer base and relationships, or have large funds to invest in A.I.
The impact on IT careers
Some developers boast online that they have not written a single line of code for some time, helped by Claude A.I.
Is this the death of the developer?
No. According to O’Brien, the role isn’t dying, but what it means to be a developer is changing, radically.
In the past we learnt coding languages, dev frameworks, needed deep IDE expertise, and had to be able to write everything from scratch.
In the future, our knowledge will likely shift to architecture & system design, A.I. orchestration & agent wrangling, reviewing and enhancing instead of creating, and will invest in prompt engineering & evaluation.
Sessions attended
May 6
- Keynote – The Evolution of Agents at MS and Everywhere
- Securing M365 Copilot Agents, Bram de Jager
- Best practices deploying Purview
- SharePoint and Data Governance, Antonio Maio
- Entra ID Management Made Easy: Learn How to Automate Access, Onboarding, Offboarding, and Movers, Julian Rasmussen
May 7
- Keynote – Building for the Future: Microsoft 365, Agents and AI
- PowerShell Meets Copilot: Feeding AI with Your Own Data, Rodrigo Pinto
- Agents with Microsoft Copilot Studio
- SharePoint Online got smarter! Organize, govern, and keep content fresh with the Knowledge Agent (AI in SharePoint)
- Staying valuable in the AI shift
- Gulp is out; Heft is in!
