BC hybrid deployments now need their Dual Use Rights key renewed every 6 months instead of once. Fair compliance move, but it's one more thing that can quietly lapse and break a deployment if nobody's tracking it.
Group Chats, podcasts, Labs, and Deep Research are getting axed from Copilot this month, plus a full app merger. This feels like Microsoft indirectly admits the sprawl wasn't working.
Flow groups let you pool 250K actions/day across up to 25 flows instead of licensing each one separately. It feels like Microsoft actually listened to us on this one instead of just adding another premium tier.
Here is the official announcement:
https://www.microsoft.com/en-us/power-platform/blog/2026/08/06/whats-new-in-power-platform-july-aug…
OpenAI-hosted models can now process your Microsoft 365 Copilot data unless your tenant was already set to “No users.” If EU Data Boundary compliance matters, check the toggle and don’t assume it’s still off.
You can check the rest about this update here: OpenAI as a subprocessor in Microsoft Online Services | Microsoft Learn
Translytical Task Flows went from preview to GA. Now you can act directly from a report instead of just viewing it and submitting a separate request. This feels like a real shift from reports being read-only to actually doing something in them.
Are you planning to build any into your reports?
The August update lets you export data from supported report visuals to Excel directly in the Power BI mobile app, with the same permissions and RLS as the Power BI service.
Is this something your team has been waiting for?
Every project I work on, the RLS setup starts simply. But it then turns into a mess of overlapping roles nobody remembers the logic for. Does anyone have a system that actually stays clean as it grows?
Has anyone fully switched over from older ADF/SSIS setups to Fabric pipelines? Or is it better to run side by side?
Everyone blames fuel and overtime for rising field service costs, but the real leak is dispatchers manually overriding the optimized schedule.
Measured it on one team, and it beat their entire fuel budget for the quarter.
Saw they simplified storage and lifecycle management in OneLake in the June release. Has anyone actually switched over, or still testing it out first?
Given the recent Azure CLI spray attacks slipping past MFA through ROPC, have you guys blocked that legacy flow in your CAPs yet?
Anyone else deal with a plugin that's worked fine for months, then just breaks after a major update with zero warning in the release notes? How are you catching this stuff before it hits production?
I have seen many agents that work great in testing but fall apart in production. Data access issues, hallucinations, and missing human oversight are the usual challenges. But every team hits a different wall depending on their stack and use case.
What's been the biggest roadblocker for your team?
Low-code makes it easy to build apps, but how is your team keeping Shadow IT under control?
The native Copilot in Dynamics works fine for basic summarization and drafting. But it falls short the moment you need it to follow specific business logic. Has anyone built a custom agent in Copilot Studio to extend it?
Dataflows worked great for reusable transformations. But Datamarts and now Fabric Lakehouses are overlapping a lot of the same use cases. Which one has your team settled on and what pushed you toward it over the others?
As Copilot agents become more autonomous, validation seems just as important as automation. How are you handling it?
Feels like every company brags about CSAT and FCR going up, but support still feels like talking to a wall.
I’d say dimensions, job queues, and account schedules. They can save a lot of manual effort, but many teams seem to overlook them.
What’s your pick?
For those who've used both, did Mythos feel noticeably more capable than Fable?