News Archives

Copenhagen, Denmark – 9 March 2026

Microsoft announced M365 E7 today, the first new enterprise license tier since E5 back in 2015. Copilot, Agent 365, Entra, and advanced security bundled into one suite at $99/user, generally available from May 1.

By Mikkel Riis

If you’ve watched Microsoft over the past couple of decades, the playbook is familiar. Nobody bundles like Microsoft. Take things customers are buying separately or still thinking about, package them so the bundle makes more sense than the parts, and let the economics do the work. It works, and it’s worked before.

But E7 does something more than the usual bundling move because it pushes three conversations that most organisations run separately into one: how to get real value from M365, how to get AI adoption to actually stick, and how to keep it all secure while you do it. That’s way more strategic than the endless discussion around licensing.

The foundation and the organisation around it are still the two biggest challenges for scaling AI, but what’s encouraging about today’s announcement is that it’s pushing these tools deeper into the platforms and workflows people already use every day.

We put out our Responsible AI paper in January, and one of the points we made was that there’s a growing gap between what the AI models can actually do and what the products and governance frameworks around them can handle.

The frontier models and what they can do today are already way ahead of what the products in users’ hands can actually deliver, and what most organisations can absorb. That gap isn’t closing because innovation keeps moving faster than both the products and the adoption around them.

Two months later, Microsoft SKU’s it all again.

Microsoft just stated that they’re running more than half a million agents internally. Think about what sits underneath that: those agents get their own identities and access rights, just like your people do. Governing that at scale is a security and identity challenge, and it’s one that looks a lot like the Zero Trust and identity governance work our security practice has been deep in for years.

Agent 365 puts structure around the same questions we’ve been working through with customers for years in our Power Platform and citizen developer work: who built this thing, what can it get to, who’s responsible when it acts on its own, and how do you keep that under control as it grows.

Copilot Cowork is worth watching too. I’ve been using Claude Cowork, Copilot, and ChatGPT Pro side by side, and the shift away from what I’d call the prompting treadmill is real. Context is king now, and the whole meta-prompting era where you spend more time engineering your prompts than doing your actual work has to end. You stop cycling through prompts, adding context, getting back slop, and start working with something that remembers what you’re doing and carries it forward across your work and daily tools.

I can build skills that are relevant to my specific work and have them automatically picked up and used across parallel work streams, which takes a lot of the busywork out of my day.

We’re seeing the same shift in software development where context-aware tools are replacing the prompt scaffolding that everyone spent the last two years building. Microsoft built Cowork with Anthropic and is bringing that into the M365 tenant, and for anyone who’s seen users try Copilot once and go back to doing things the old way, this could be what changes that. But it also means the adoption and governance work we do with customers just got even more important.

Microsoft just stated that they’re running more than half a million agents internally. Think about what sits underneath that: those agents get their own identities and access rights, just like your people do. Governing that at scale is a security and identity challenge, and it’s one that looks a lot like the Zero Trust and identity governance work our security practice has been deep in for years.

The folks who will move quickly on all of this are the ones who are ready to meet what E7 assumes is already in place. That’s a challenge we’ve been carrying for a while, and E7 just shines a brighter light on it, the same way Copilot exposed every gap in security, sensitivity labelling, and data hygiene that organisations had been living with.

The foundation and the organisation around it are still the two biggest challenges for scaling AI, but what’s encouraging about today’s announcement is that it’s pushing these tools deeper into the platforms and workflows people already use every day. I’m convinced that will help speed things up.

Reach out to us if you want to discuss. Me and my colleagues are always happy to share what we’re seeing and how we can help.

Read more about the topic: