SIGNALS HUB
View for
Back to Home
About Signals Hub

One place to follow what's changing in AI at work

Signals Hub is a personal project. It joins five public Microsoft sources into one record per feature, so you can see what is coming, what shipped and what needs a decision without reading five feeds.

5Sources, joined on Microsoft's own IDs
DailyEvery source refreshed each morning
FreeNo sign-in, no cookies

Five sources

Roadmap
Microsoft’s AI at Work Roadmap (formerly the Microsoft 365 Roadmap): what is planned, its status and its target months. Answers “what is coming, and when?”
Message Center
The posts admins get in their tenant: rollout dates per ring and cloud, act-by dates and retirements. Answers “what changes in my tenant, and do I need to act?” The posts come from one tenant, so yours may have more or fewer.
Release notes
The Microsoft 365 Copilot release notes: what actually shipped, and something to try. Answers “is it really here?”
Blogs
Microsoft and community blogs that cite roadmap or Message Center IDs. Answers “what are people saying about it?”
GitHub
GitHub Copilot admin changes. Coming next: they share no ID with the other sources yet.

Two layers

Data layer
Joins the sources on Microsoft’s own IDs and keeps their history: Roadmap IDs found in posts, release notes and blogs, or the Message Center ID when there is no roadmap item. Every version of a post and every status and date move on the roadmap is kept, and each feature is sorted into Act now, Try it now, Get ready, Review later or Check it’s live.
AI layer
Reads every visible Message Center post the way an admin or adoption lead would: rollout dates per ring and cloud and whether they moved, impact on daily work from None to High, who it affects, the kind of work, the AI area, and whether it turns on by default. It also writes two short summaries of each post in its own words, one for adoption leads and one for admins, and a note on what changed when Microsoft updated the post. Each result is checked before it is used: rollout dates must be quoted from the post, and summaries may not copy the post, contain links or name dates the post does not give. Lists show only these facts and a one-sentence reason; the summaries are in the Message Center tab of an opened feature.

Two views

Every page has a View for switch. It changes what leads, not what is there: both views see every feature.

Adoption leads
Leads with what is worth telling users: shipped or rolling out now, with a real effect on daily work, and something to try.
Admins
Leads with what needs a decision: act-by dates, changes that turn on by default, retirements and new controls for AI and agents.

Update times

Each source is read once a day, in this order.

04:00 UTC
Roadmap
05:00 UTC
Message Center
06:00 UTC
Blogs
07:00 UTC
Release notes
08:00 UTC
AI layer: new and changed Message Center posts

What an opened feature shows

Open any row in Feature Lookup to see what each source says about it, in tabs. Only the tabs a feature has appear.

In the lists
Facts, dates, IDs and the AI layer’s one-sentence reason. Microsoft’s Message Center text never appears on the site.
Message Center
Opening a feature shows every visible post linked to it, newest first: title, MC ID, services, tags and dates, then an AI summary for the current view (the other view’s is one click away) and what changed in the latest update. Earlier versions show their dates and what changed, never their text. The summaries are written by AI from posts received in a test tenant; to read a post itself, look it up by its MC ID in Message Center in your own Microsoft 365 admin center. Hidden posts never appear.
Roadmap
The full roadmap item: description, status, products, platforms, release phases, clouds, preview and GA months, links, and every status and month change recorded since tracking began.
Release notes
Every linked release-note entry with all of its sections and learn links, credited under CC BY 4.0.
Blogs
Each linked article’s title, feed, date and the summary from its RSS feed (not the article itself), with a link to read it.
AI evaluation
What the AI layer read from each post: rollouts per ring and cloud with their dates, impact, direction, audience, work type, AI area, default state and its reason. The quotes it checks dates against stay internal.

Tell me what you think

What's useful, what's missing, what you'd change. I'd love to hear it. Send me a message or comment on LinkedIn.