Yes. Microsoft has an active Exchange Online outage today that would cause exactly this symptom in Outlook Classic.

The outage

• Microsoft’s message: “We’re investigating reports of users experiencing issues accessing Exchange Online mailboxes through multiple connection methods in the Americas. More details can be found in the Microsoft 365 admin center under EX1550543.”
• Start time: November 12, 2026, 4:17 PM UTC, which is 8:17 AM Pacific.
• Affected connection methods: Outlook on the web, MAPI, Exchange ActiveSync (EAS), and the REST API. Outlook Classic connects to Microsoft 365 over MAPI, so it is affected.
• Region: North America, Latin America, Canada, and Chile
• Status: Microsoft reports connectivity errors on a portion of the Exchange Online infrastructure that receives and processes access requests. No resolution time is given. The next update is due by Thursday, November 12, 2026, at 8:00 PM UTC, which is 12:00 PM Pacific.
• Downdetector: Reports surged at around noon and climbed rapidly, with over 12,000 reports within a half hour (noon Eastern is 9:00 AM Pacific).

The office that blamed last night’s Windows Updates ➜ see Note 1 below

Note 1: Most likely the restart is a coincidence and the outage is the cause. The outage began at 8:17 AM Pacific this morning, when that office would have started using Outlook. November’s Patch Tuesday was November 10, and Windows 11 gets KB5132061 with no known issues listed. To confirm, check whether Outlook on the web (outlook.office.com) also fails for that office. If it does, the problem is the outage, not their PCs. Don’t rebuild Outlook profiles or roll back updates until Microsoft resolves EX1550543.

Separately, the one actively exploited bug fixed this month is in Outlook (CVE-2026-124194). That fix doesn’t come through Windows Update at all, it comes through Office’s own updater. Once the outage is resolved, it’s worth making sure Office is up to date on both clients’ machines.

Sources: KHQ – Outlook outage; StatusStack – Exchange Online status (mirrors Microsoft’s status page); PIX11 – Thousands report outage with Microsoft Outlook and 365; Digital Citizen – November 2026 Patch Tuesday; The Tech Hacker – November 2026 Patch Tuesday.

Current status (checked 10:55 AM Pacific): Microsoft Exchange Online incident EX1550543 is still open. Its status is “Monitoring” (minor). It is not resolved. Microsoft’s next update is due at 12:00 PM Pacific.

Ways to find out when it is fixed

1. Microsoft 365 admin center Service Health. Use this if you have admin access to either client’s tenant. Go to admin.microsoft.com → Health → Service health and search for EX1550543. This is the most reliable source.
2. Email alerts from Service Health. On the same Service health page, click Customize, open the Email tab, and turn on incident notifications for Exchange Online. Microsoft then emails you each update, including the one that says the incident is resolved. You have to set this up in each tenant.
3. Public sources, no login needed: status.cloud.microsoft, and the @MSFT365Status account on X. Both report only large incidents.
4. I can check for you. I can set up a scheduled task that checks the incident status at intervals (for example, every hour) and sends a notification to your phone when it shows resolved. ➜ see Note 1 below

Will Outlook recover on its own? You’re partly right that we don’t know for certain. Usually, though, Outlook Classic keeps retrying the connection, and the exclamation mark clears by itself within a few minutes of Microsoft’s fix. The update so far says nothing about users needing to restart Outlook or any other client. Microsoft’s final “resolved” message normally says if clients need restarting. If it doesn’t, here’s what I’d tell clients:

• Wait about 15 minutes after Microsoft says it’s fixed.
• If the exclamation mark is still there, close Outlook and make sure OUTLOOK.EXE has ended in Task Manager, then reopen it.
• Restart the computer only if reopening Outlook doesn’t fix it.

Note 1: A scheduled task runs in a new session and sends you a push notification. Tell me how often to check and when to stop, for example “every hour until 6:00 PM Pacific”. I’ll check once more myself after Microsoft’s 12:00 PM Pacific update if you ask.

The checks are scheduled. There are 12 of them, one every 5 minutes from 12:00 PM to 12:55 PM Pacific today, and nothing runs after the 12:55 PM check.

How the checks work

• Each check reads Microsoft’s latest update on Exchange Online incident EX1550543 (the Microsoft outage stopping your clients’ Outlook from connecting). It compares that with the last update already seen in this conversation, the 10:55 AM “collected networking logs” update.
• A push notification is sent only when Microsoft has posted something new. If nothing has changed, a check sends no notification and adds only a one-line note to this conversation.
• If Microsoft marks the incident resolved, that check notifies you and deletes the remaining checks. The notification starts with “RESOLVED” and says whether Microsoft mentions restarting Outlook.
• The 12:00 PM check also picks up any update Microsoft posts between now and 12:00 PM.

Limits ➜ see Note 1 below

Note 1:

• Possible delay: The checks read third-party sites that copy Microsoft’s status page (statusstack.com, isdown.app and statusfield.com), not Microsoft’s own page. I can’t read that without being signed in to a tenant. Each copy lags Microsoft by a few minutes, and my page reader reuses a saved copy of a page for 15 minutes. That’s why the checks take turns among the three sites. Even so, a notification can arrive several minutes after Microsoft posts its update.
• Phone delivery: If you’re looking at this conversation when a check finds something new, the phone notification is skipped and the update just appears here.

The 12:00 PM check (1 of 12) found a new Microsoft update on incident EX1550543, the Exchange Online outage affecting your clients’ Outlook. It is not resolved yet. I sent the push notification to your phone.

What Microsoft says now:

• Microsoft has identified a pattern in which affected servers are becoming intermittently unavailable, which strains the remaining infrastructure. It has begun spreading Outlook connection traffic across its wider worldwide infrastructure and is reviewing recent changes to find the source of the increased traffic. Some latency may result.
• A later statement quoted in the news adds: “We’ve paused all deployments, we’re moving additional traffic off the affected infrastructure, and we’re restarting affected servers to return them to service.”
• Status: Monitoring (ongoing)
• Restarting Outlook: Not mentioned.
• Next update: 2026-11-12 20:30 UTC, which is 12:30 PM Pacific.

The remaining 11 checks (12:05 PM to 12:55 PM Pacific) are still scheduled. They will notify you only when Microsoft posts something newer than this update.

Sources: StatusStack – Exchange Online status; WGCU – Outage reported nationwide for Outlook; WindowsForum – Exchange Online outage.

The 12:05 PM check (2 of 12) found nothing newer than the 12:00 PM update on incident EX1550543, so I sent no notification. Microsoft’s next update is due at 12:30 PM Pacific.

The 12:10 PM check (3 of 12) found nothing newer than the 12:00 PM update on incident EX1550543, so I sent no notification. News reports say Microsoft has not provided an estimated time for when Outlook services will be fully restored.

The 12:15 PM check (4 of 12) found a new Microsoft update on incident EX1550543, the Exchange Online outage affecting your clients’ Outlook. It is not resolved yet. I sent the push notification to your phone.

New from Microsoft (Microsoft 365 Status on X):

• Cause narrowed down: the elevated load isn’t driven by an increase in traffic, but by a number of servers being unexpectedly unavailable, straining the remaining infrastructure.
• Fix in progress: Microsoft is testing a configuration change to address a problem that’s causing servers to fail automated health checks and become unavailable. In parallel, it has initiated a restart of all affected servers and is monitoring health as they return to service.
• Microsoft now also tracks a second incident ID, MO1550821. Updates are posted in the admin center as well as status.cloud.microsoft if the admin center is not available to you.
• Restarting Outlook or the computer is still not mentioned.

The remaining 8 checks (12:20 PM to 12:55 PM Pacific) are still scheduled.

Sources: Microsoft 365 Status on X (two posts).

The 12:20 PM check (5 of 12) found nothing newer than the 12:15 PM update (Microsoft testing a configuration fix and restarting all affected servers) on incident EX1550543, so I sent no notification.

The 12:25 PM check (6 of 12) found a new Microsoft update on incident EX1550543, the Exchange Online outage affecting your clients’ Outlook. Microsoft posted it at 12:23 PM Pacific. The incident is not resolved yet. I sent the push notification to your phone.

New from Microsoft:

• Microsoft says it hasn’t “identified any recent change correlated with the start of impact.” Its investigation suggests the failed health checks may come from “a specific component” that processes server health. It has taken steps to stop more servers being affected while it returns the affected servers to service.
• Restarting Outlook: The update doesn’t tell users to restart Outlook or take any other action.
• Signs of recovery: Downdetector reports fell from a peak of about 130,000 to about 77,000 by 19:41 UTC, which is 11:41 AM Pacific. The live blog says the outage isn’t over.

The remaining 6 checks (12:30 PM to 12:55 PM Pacific) are still scheduled.

Source: Tom’s Guide – live updates on the Outlook outage.

The 12:30 PM check (7 of 12) found nothing newer than Microsoft’s 12:23 PM update (failed health checks traced to a specific server-health component) on incident EX1550543, so I sent no notification. Microsoft’s latest post is still the “specific component” update, shown on X as 5m old.

The 12:35 PM check (8 of 12) found nothing newer than Microsoft’s 12:23 PM update on incident EX1550543, so I sent no notification.

The 12:40 PM check (9 of 12) found nothing newer than Microsoft’s 12:23 PM update on incident EX1550543, so I sent no notification. TechRadar’s live blog reposted that same update at 12:35 PM, and its author notes “We’re waiting for a solution.”

The 12:45 PM check (10 of 12) found a new Microsoft update on incident EX1550543, the Exchange Online outage affecting your clients’ Outlook. Microsoft posted it at 12:37 PM Pacific. The incident is not resolved yet. I sent the push notification to your phone.

New from Microsoft:

• Microsoft is deploying a targeted change to bypass a component on some servers to test whether that restores availability. It is also investigating why affected servers show high processor usage.
• Status: No update says services are recovering or resolved. The latest update describes an ongoing investigation and a test fix.
• Restarting Outlook: Microsoft has not mentioned it. The only restarts described are Microsoft restarting affected servers.

Two checks remain, at 12:50 PM and 12:55 PM Pacific. If the incident is still open after 12:55 PM, tell me if you want more checks scheduled.

Source: Tom’s Guide – live updates on the Outlook outage.

The 12:50 PM check (11 of 12) found nothing newer than Microsoft’s 12:37 PM update (testing a change that bypasses the faulty component on some servers) on incident EX1550543, so I sent no notification. One check remains, at 12:55 PM Pacific.

The 12:55 PM check (12 of 12, the final check) found nothing newer than Microsoft’s 12:37 PM update (testing a change that bypasses the faulty component on some servers), so I sent no notification. Incident EX1550543 is still open. Both remain open with no end time, and Microsoft has not posted a customer-visible recovery. All scheduled checks have now finished. If you want them continued, tell me how often and until what time.

Source: StatusStack – Microsoft 365 and Exchange Online outage, November 12, 2026.

The first check is scheduled for 2:05 PM Pacific today. Microsoft promised its next update on incident EX1550543 (the Exchange Online outage) by 5 p.m. ET, which is 2:00 PM Pacific.

How the checks will run

• Every check sends a push notification to your phone. It starts with RESOLVED, NEW UPDATE or NO NEW UPDATE, says whether Microsoft mentions restarting Outlook, and gives the time of the next check.
• Each check schedules the next one. If Microsoft has published a next-update time, the next check runs 5 minutes after it. If not, the next check runs 1 hour later.
• Checks stop when Microsoft marks the incident resolved. ➜ see Note 1 below

Since the 12:55 PM check, Microsoft has posted two updates. The service is recovering but is not confirmed resolved:

• 12:49 PM Pacific: While testing the fix, Microsoft found many affected servers had become unresponsive. It is restarting them and applying the same approach to more infrastructure.
• 1:30 PM Pacific: Microsoft validated an alternative health check method on some affected servers and is applying it more broadly. It is also preparing automated recovery to restart unresponsive servers.
• Downdetector: reports fell from about 130,000 at peak to over 50,000.
• Microsoft still has not said users need to restart Outlook.

Note 1: You did not give an end time, so the checks keep running, through the evening if needed, until Microsoft marks the incident resolved. Tell me if you want them to stop at a set time instead.

Sources: WebProNews – Microsoft 365 outage cuts off Outlook for hours; Tom’s Guide – live updates on the Outlook outage.

I updated the scheduled check for incident EX1550543 (the Exchange Online outage). Checks will not run between 8:00 PM and 7:00 AM Pacific. If a check would land in that window, it is moved to 7:00 AM Pacific the next morning. The 7:00 AM check will report anything Microsoft posted overnight. Everything else is unchanged: the first check runs at 2:05 PM Pacific today, every check sends a push notification to your phone, and checks stop when Microsoft marks the incident resolved.

The 2:06 PM check found no new update on incident EX1550543 (the Exchange Online outage). I sent the push notification to your phone.

• Latest Microsoft update (still the 1:30 PM Pacific one): “We’ve validated an alternative health check method on a subset of affected infrastructure…” It says the change is being applied more broadly and an automated recovery process is being prepared.
• Status: Recovering, not resolved. The blog says there are still 50,000+ Outlook reports and “we’re not in the clear yet.”
• Restarting Outlook: Microsoft has not mentioned it.
• Next check: Microsoft has not published a next-update time, so the next check runs one hour from now, at 3:06 PM Pacific.

Source: Tom’s Guide – live updates on the Outlook outage.

The 3:09 PM check found a new Microsoft update on incident EX1550543 (the Exchange Online outage). It is a setback: the outage is not resolved. I sent the push notification to your phone.

• New from Microsoft (posted about 2:26 PM Pacific): “We’re continuing to pursue all available pathways to restore service,” and the actions taken so far “have not provided the anticipated relief.” Microsoft says a core process on affected servers is becoming overwhelmed. It is isolating part of its infrastructure, restarting components, and gradually reintroducing traffic. Microsoft also said it was assessing whether reverting a small number of recent configuration updates it identified could help accelerate recovery.
• Signs of improvement: Downdetector reports had fallen to about 40,000 from a peak above 131,000, but access was still not restored for most users. Tom’s Guide also reports that some users said Outlook was back up.
• Restarting Outlook: Microsoft has not mentioned it.
• Next check: Microsoft has not published a next-update time, so the next check is at 4:09 PM Pacific.

Sources: TechRadar – live updates on the Outlook outage; Tom’s Guide – live updates on the Outlook outage; Interesting Engineering – Microsoft 365 outage.

The 4:10 PM check found a new Microsoft update on incident EX1550543 (the Exchange Online outage). Recovery has started but the outage is not resolved. I sent the push notification to your phone.

• New from Microsoft (about 3:05 PM Pacific): Microsoft says it is returning some affected servers to service, which “remain stable so far.” It is also shifting traffic to alternate routes and “observing limited recovery.” Users may see email arrive in pieces as a backlog clears.
• Restarting Outlook: Not needed so far. Microsoft says users “may experience a gradual recovery as their clients reconnect.” This answers your earlier question: Microsoft expects Outlook to reconnect by itself. Tell clients that mail may arrive in batches.
• Downdetector: Reports were “just over 37,000” at 5:41 pm ET, and “Some users are reporting that Outlook is back up.”
• Next check: Microsoft has not published a new next-update time, so the next check is at 5:10 PM Pacific.

Sources: TechRadar – live updates on the Outlook outage; Tom’s Guide – live updates on the Outlook outage.

The 5:10 PM check found a new Microsoft update on incident EX1550543 (the Exchange Online outage). It is a setback, and the outage is not resolved. I sent the push notification to your phone.

• New from Microsoft: “The servers we returned to service can become unresponsive again once they receive traffic; we’re analyzing diagnostic data to determine what’s triggering the behavior. In parallel, we’re temporarily limiting which types of traffic can reach the affected infrastructure in some locations.”
• What this means for your clients: The partial recovery Microsoft reported around 3:05 PM may not hold. Clients whose Outlook reconnected could lose the connection again.
• Restarting Outlook: Microsoft has not mentioned it.
• Next check: Microsoft has not published a next-update time, so the next check is at 6:11 PM Pacific.

Source: Microsoft 365 Status on X.

The 6:11 PM check found a minor new Microsoft update on incident EX1550543 (the Exchange Online outage). The outage is not resolved. I sent the push notification to your phone.

• New from Microsoft: We’ve reviewed known impact scenarios and revised the list of impacted services and features, which can be viewed in the “more info” section of the communication MO1550821 in the Microsoft 365 admin center. (MO1550821 is Microsoft’s second incident ID for the wider Microsoft 365 impact.)
• Not in my 5:11 PM report: The 4 PM Pacific update also said: “Additionally, we’re rebuilding a subset of affected servers using a previous configuration to validate if this prevents them from becoming unresponsive.”
• Restarting Outlook: Microsoft has not mentioned it.
• Next check: Microsoft has not published a next-update time, so the next check is at 7:12 PM Pacific. That is the last check tonight. If the 7:12 PM check also finds no next-update time, the following check is at 7:00 AM Pacific tomorrow.

Sources: Microsoft 365 Status on X; Fox Business – Microsoft 365 connectivity issues.

The 7:13 PM check found a new, positive Microsoft update on incident EX1550543 (the Exchange Online outage). It is still not resolved. I sent the push notification to your phone.

• New from Microsoft: “We’re seeing positive results with our efforts to restart affected infrastructure and recover server availability in an isolated manner. We’re actively monitoring the performance of these portions of infrastructure to verify stability”.
• Restarting Outlook: Microsoft has not mentioned it.
• Next check: No new next-update time is published, so the next check would fall after 8:00 PM. Checks are paused until 7:00 AM Pacific on Friday, November 13. That check will report everything Microsoft posts overnight.

Source: Microsoft 365 Status on X.

Yes, I have your instructions for a YouTube-safe replay. They say to:

• produce it as an HTML file. Your website publisher doesn’t exist yet, so I’ll send you the file.
• include every prompt and every response in order, each with its date, time and response number.
• replace any identifying names with coded designators. This chat has none: the clients were never named. ➜ see Note 1 below
• replace raw tool output with a one-line note of what was done.
• add a substitution key at the end.
• open with the AI, the model and the effort level.

I’ll use “Let AI Build Your Microsoft Outage Alert System (Phone Notifications Included)” as the title.

Before I build it, I need one answer: what effort level should the introduction show? The model is Claude Opus 5.5. Your instructions say to ask when the effort level isn’t given.

The 7:00 AM check on incident EX1550543 has not run. EX1550543 is the Exchange Online outage. The check is still waiting, with no run recorded. It is tied to your computer, which was probably off or asleep at 7:00 AM, and it may run as soon as that computer is back online. ➜ see Note 2 below

Note 1: The chat names you (Doug) and Microsoft. You’re the presenter, and Microsoft is the subject of the show, so I’ll leave both as they are unless you say otherwise.

Note 2: Do you want me to run the outage check now instead? If you say yes, I’ll delete the waiting 7:00 AM check so you don’t get a duplicate notification.

Outage check (incident EX1550543, the Exchange Online outage): I ran the check and sent a push notification to your phone. I deleted the 7:00 AM check that never ran. The incident is recovering but not resolved.

• Latest: As of November 13, 2026, 12:30 UTC, Microsoft’s latest update reported progress but warned that intermittent issues may continue.
• Delayed mail: Microsoft extends the timeout window so queued mail keeps retrying instead of failing, and prioritizes backlog processing. The window is now 48 hours, so mail sent to your clients during the outage should arrive late instead of bouncing.
• Overnight: Partial recovery in some regions, but not sustained. Microsoft then scales up VMs to handle the traffic surge. Some machines are stuck in a bad state.
• Restarting Outlook: Microsoft has still not mentioned it.
• Next check: 8:05 AM Pacific. Microsoft’s latest update promised the next one by Friday, November 13, 2026, at 4:00 PM UTC, which is 8:00 AM Pacific.

Why the 7:00 AM check didn’t run: You’re right that your computer wasn’t asleep, so I was wrong to suggest it. The 7:00 AM check shows no run recorded, and I can’t tell why. ➜ see Note 1 below

The presentation: The file “Let AI Build Your Microsoft Outage Alert System 2026-11-13.html” is attached. The introduction reads “Claude desktop chat, Opus 5.5 model, effort level High.” The page has a short overview, then every prompt and every response from #1 to #25, with each scheduled-check trigger marked as an automatic message. Your name and Microsoft’s are kept, as you approved. No other substitutions were needed.

Note 1: The 8:05 AM check is set up the same way as the 7:00 AM one. If you don’t get a notification by about 8:10 AM, tell me and I’ll run the check by hand.

Sources: SOCRadar – Microsoft Outlook outage; LazyAdmin – Microsoft 365 outage status; Cloud Admin Hub – EX1550543.

They’re in Claude’s saved memory for this Project (“Client Computer Support Work”), in a file of your preferences for this Project. They aren’t in the Project instructions or in any of the five Project docs.

That file holds your YouTube-safe rules: the replay format, coded designators, HTML output, naming the AI, model and effort level, and the two-time-stamp handling. It also holds your other Project preferences, such as the response header and the PowerShell .ps1 plus .bat delivery. It was saved from earlier chats in this Project, and it was updated again this morning.

1. Notification on every check: Yes, you recall correctly, and it is set up that way. The 8:05 AM check on incident EX1550543 (the Exchange Online outage) is due to run now. Its instructions say to send a push notification every time, starting with RESOLVED, NEW UPDATE or NO NEW UPDATE. Every check it schedules after that carries the same instruction. ➜ see Note 1 below

2. AI designation for replays: Yes. Your saved preferences for this Project now say the replay introduction must name the AI, the surface (Claude desktop chat or Claude Code), the model and the effort level, for example “Claude desktop chat, Opus 5.5, effort level High.” The HTML file from Response #25 already uses that wording.

3. Handoff prompt: Start the new chat inside this same Project (“Client Computer Support Work”) so it can use your Project preferences. Before you send the prompt, add your scripts folder as a connected folder in the desktop app. You can’t download .bat files from Claude, so the new chat needs that folder to write the files there. Paste this:

New task: build a publisher that posts my YouTube-safe HTML replays to my website.

Background
- In this Project, Claude produces "YouTube-safe replays" of chat sessions as self-contained HTML files (inline CSS, no external assets). The first one is "Let AI Build Your Microsoft Outage Alert System 2026-11-13.html". I present these live on my YouTube stream from my website, and I'm building an ongoing collection of examples of using AI to automate tasks for clients.
- I already have a working Python script that publishes blog posts about my live stream shows to my WordPress site. I will attach that script (plus its .bat launcher and a description of its config) as resource material. Use it as the model for authentication, posting and logging. Do not ask me to rebuild what already works.

What I want
1. Read my existing publisher script first and tell me how it connects, authenticates and posts.
2. Propose how an HTML replay should appear on the site (post vs page, category, tags, draft vs publish, and how to keep the inline CSS from being stripped by WordPress). Recommend one approach and ask me before building.
3. Propose file names for the new script set and wait for my approval before writing code.
4. Build the publisher: a .py script plus a .bat launcher, written into my connected folder.

My code conventions
- Full files, not snippets. Every script gets a metadata header: script name, language, environment, objective, version, date and local time, revision summary.
- File names use spaces, not underscores. Related .py/.bat/.txt files share one base name, with a phase suffix (P1, P2) for multi-phase work.
- Relative paths from the script's own folder. Every Python script writes its own .txt log, with a YYYY-MM-DD-HHMM timestamp on log and output files.
- The .bat launcher starts with: call "%USERPROFILE%\miniconda3\Scripts\activate.bat" base
- No placeholder values in code. Ask me for anything missing.

Safety
- This chat may be shown on my YouTube channel. Never put passwords, API keys, application passwords or real client names into replies or code. Keep secrets in a local config file or environment variable that is never displayed. Warn me right away if I paste anything identifiable.

Note 1: The scheduled 7:00 AM check never ran, for a reason I couldn’t find, and the 8:05 AM check is set up the same way. If no notification reaches your phone by about 8:10 AM, tell me and I’ll run the check by hand. Also, the notification tool can skip the phone if you’re looking at this chat when a check runs. In that case the result appears here instead.

Phone notifications: Yes, they go through the Claude app on your phone. I send each notification through Claude’s own notification system, and it arrives as a Claude app notification on your phone. No text message or email is involved.

Adding the scripts folder: Add it in the new chat, inside this Project. Connected folders belong to each chat, not to a Project. The Project limit you’re thinking of applies only to memory: a Project’s memory is kept separate from your other chats. It doesn’t affect connected folders. Two ways to add the folder in the new chat:

• Click the “Add folder” button in the Claude desktop app, on the computer that has your F: drive.
• Or tell the new chat the folder’s path, and it will show you a permission prompt to approve.

The 8:05 AM check did not run either, so the scheduled checks are not working right now. ➜ see Note 1 below

• I ran the check by hand at 8:10 AM and sent the push notification. Microsoft’s newest update says: “We’re observing sustained improvements in service availability as recovery progresses across affected Microsoft 365 services.” Incident EX1550543 is still not resolved, and Microsoft has not mentioned restarting Outlook.
• I deleted the 8:05 AM check that never ran, so it can’t fire late and duplicate the notification. No check is scheduled now.

Note 1: Thursday’s checks all ran. The 7:00 AM and 8:05 AM checks today never ran. All of these checks are tied to the computer this chat was linked to on Thursday. If you’re on a different one of your computers this morning, that may be the reason, but I can’t confirm it. A more reliable option is a scheduled task that runs in the cloud and isn’t tied to any computer. It would check hourly from 7:00 AM to 8:00 PM and send a phone notification every time. Do you want me to set that up, or use the same kind of check as before?

Source: Cloud Admin Hub – EX1550543.

Cloud scheduled task: set up. I created “EX1550543 outage check (hourly).” It runs in the cloud and isn’t tied to any of your computers.

• Schedule: every hour at 5 minutes past, from 7:05 AM to 7:05 PM Pacific, every day. The first run is at 9:05 AM today.
• Notification: a push notification to the Claude app on your phone after every run, starting with RESOLVED, NEW UPDATE or NO NEW UPDATE, and saying whether Microsoft mentions restarting Outlook.
• Stopping: when a run finds the incident resolved, it turns the task off itself.
• Approvals: the task runs automatically, without asking for approval. You can change that in the task’s settings.
• Where the results appear: each run starts its own new session, so results come to your phone and to that session, not into this chat. ➜ see Note 1 below

The earlier checks were not running in the cloud. Each one was a reminder delivered back into this chat. All of them were tied to one specific computer: the one this chat was linked to on Thursday. You’re right that sitting at that computer doesn’t matter. What mattered was that one computer, and the Claude desktop app on it being open and connected. Having your other two computers on didn’t help. Thursday’s checks ran. This morning’s 7:00 AM and 8:05 AM checks were never started, and I can’t see why. The new task avoids that dependency.

YouTube-safe replays: Agreed. Every replay will include the whole chat: the outage work, finding out the scheduled checks weren’t running, the website-posting discussion, and the replay discussion itself, up to and including your latest prompt. The file I made in Response #25 stops at #25, so I’ll rebuild it whenever you ask, to include everything through that point.

Note 1: The last run each day is at 7:05 PM, not 8:00 PM. The task sets times by the hour, and 7:05 PM is the last one before your 8:00 PM cutoff.

I saved the updated replay as “Let AI Build Your Microsoft Outage Alert System 2026-11-13.html” in your Pending_Articles folder. That folder was empty before.

It now covers every prompt and response from #1 through #30:

• Friday morning: the checks that never ran, the move to the cloud task, the handoff prompt, and your request to include the whole chat.
• Response #25 is now shown in full, replacing the short placeholder in the first version.
• The overview has two new steps about the failed checks and the cloud task. The notes say the off-topic parts are included at your request.
• The only substitutions are your name and Microsoft’s, both kept as you approved. Folder paths and device names appeared only in tool output, so they don’t appear in the file.

The cloud scheduled task “EX1550543 outage check (hourly)” is working. Its record shows:

• Last run: started 2:06 PM Pacific, finished 2:07 PM, status succeeded.
• Next run: 3:05 PM Pacific.
• Phone notifications: turned on for this task. The record doesn’t show whether a notification was delivered, but the notifications you’re getting a few minutes after each hour match the run times. Each run starts a minute or two after :05 and takes about a minute.
• State: still on. A run hasn’t found the incident resolved yet, so the task hasn’t turned itself off.

I can only see the most recent run in the record, not a full history of today’s runs. That record doesn’t include what the run found, so the notifications on your phone are where you’ll see each result.

YouTube-safe flag: The scheduled-task list I pulled for this check also returned your separate “Daily backup notice check” task. Its instructions contain client names and email addresses. That text appeared only in raw tool output, not in this reply. Replays leave out raw tool output, so it won’t appear in the HTML.

No, I wasn’t able to find the cause.

What the records showed:

• Thursday’s checks ran on time, each recorded as fired and succeeded.
• The Friday 7:00 AM check and the Friday 8:05 AM check both stayed in “waiting” past their start times. Neither showed a run, a failure or an error reason. They were never started at all, so nothing ran and then failed.
• All of these checks were one-time reminders tied to the same computer through the Claude desktop app. The two Friday checks were set up the same way as Thursday’s.

What I can’t see: the system that starts scheduled tasks doesn’t show me why it skipped a check. I deleted both Friday checks to prevent late duplicate notifications, so their records can’t be examined now.

Possible causes I can’t confirm: the link between this chat and that computer’s Claude desktop app may have dropped overnight, or there may have been a problem on the scheduling service’s side.

The cloud task avoids both, because it doesn’t depend on any computer or on this chat.

It is one task. “EX1550543 outage check (hourly)” is a single recurring scheduled task. It does not create new tasks. Each run opens a fresh working session to do its check, but the task itself stays the same and keeps its schedule until a run finds the incident resolved and turns it off. That replaced the earlier approach, where each check created the next one.

Overnight: changed. The task now runs every hour at 5 minutes past, around the clock, starting with the 3:05 PM run today. Before the change it ran only from 7:05 AM to 7:05 PM.

You’re right about nighttime Focus on an iPhone. It stops notifications from making a sound or lighting the screen, and they wait silently in Notification Center until morning. ➜ see Note 1 below

Note 1: Focus lets a notification through in two cases:

• The Claude app is on that Focus’s list of allowed apps.
• “Time Sensitive Notifications” is turned on for that Focus and a notification is marked time-sensitive.

If Claude isn’t on the allowed list, the overnight runs will stay silent. Tell me if you’d rather go back to 7:05 AM to 7:05 PM, and I’ll change the schedule back.

I agree: with the Claude app not allowed in your Sleep Focus and Time Sensitive Notifications turned off, overnight notifications will arrive silently. The task stays as it is, hourly at 5 minutes past, around the clock.

Viewing results in the desktop app: You can’t see a list of the phone notifications in the desktop app, but you can see what each check found, which is the same information. Scheduled tasks run remotely, so they run on their cadence even when your computer is asleep or the Claude Desktop app is closed. Review upcoming and past runs by clicking “Scheduled” in the left sidebar on any surface.

• Click “Scheduled,” then “EX1550543 outage check (hourly),” to see its list of past and upcoming runs.
• Click any run to open it as a full session. From there you can see what Claude did. Each run ends with a 2 to 5 line summary: the label (RESOLVED, NEW UPDATE or NO NEW UPDATE), Microsoft’s latest update, and whether Microsoft mentions restarting Outlook.
• “On any surface” means the same list is in the Claude app on your phone, so you can check it there too.

The phone notification itself is only a short alert. The full result of each check stays in that run’s session.

Sources: Schedule recurring tasks in Claude Cowork – Claude Help Center; Scheduled tasks on the web – Claude Code Docs.

Yes, renaming is safe. The task runs on its internal ID and its schedule, not its name. Nothing in its instructions uses the name, and the notification text doesn’t include it, so renaming won’t change when it runs or what it sends.

In that menu:

• Rename is safe.
• View scheduled task is safe. It opens the task’s details and run history.
• Delete: don’t use it. Deleting a scheduled task stops all future checks and notifications.
• Archive: avoid it too. I can’t confirm whether archiving that item affects the task.

If you’d rather, tell me the new name and I’ll rename it from here.

The scheduled task itself is unaffected. I checked it:

• Name: still “EX1550543 outage check (hourly)”. The rename did not reach the task.
• State: still on.
• 3:05 PM run: started 3:05 PM, finished 3:06 PM, succeeded.
• Next run: 4:05 PM Pacific.

What you renamed: the sidebar item for one run’s session, most likely the 3:05 PM check, not the task. Each run opens its own session, and the app lists those sessions in your sidebar. The dot next to “MS Outlook scheduled outage ch…” marks that session as unread, meaning it holds a result you haven’t opened.

The lightning bolt and the indent: I can’t confirm exactly what the app uses them for. Most likely they mark a session created by a scheduled task, and possibly one that was still running at the time. Renaming seems to have turned it into a regular chat in the list, without that marker. Either way, the change affects only how that one session appears in your sidebar. It doesn’t change the task, its schedule or its notifications.

To rename the task itself, open it with “View scheduled task,” or tell me the name you want and I’ll change it from here.

Response #37 produced this updated replay, covering every prompt and response from #1 through #37, and saved it to Doug’s pending articles folder.