CIA Brief 20261121

image

Security

Microsoft 365 & Windows

  • The Great Exchange Online Meltdown of November 2026
    Tony Redmond goes through the Exchange Online outage: Microsoft’s preliminary post-incident report puts EX1550543 at 23 hours and 50 minutes, with a separate incident (MO1550821) covering the knock-on effects on other Microsoft 365 services. It’s a good, clear account to share with clients who want to know what happened. It’s also a reminder to agree how you’ll communicate with clients before the next big outage.
    https://office365itpros.com/2026/11/16/exchange-online-outage-nov26/
  • Copilot at Microsoft Ignite: Expanding what’s possible with intelligence built for work
    This is the main Copilot announcement from Ignite. Home (Chat and Cowork together), Office in Copilot and Copilot Managed Runtime are now GA, Copilot Code is coming in the next few weeks, and Autopilot is starting to roll out to Frontier customers. Auto routing now picks the model and effort for each task, so expect more ‘which model is it using?’ questions from clients.
    https://www.microsoft.com/en-us/copilot/blog/2026/11/17/copilot-at-microsoft-ignite-expanding-whats-possible-with-intelligence-built-for-work/
  • Office at Ignite 2026: From apps to agentic system of work
    Office in the Copilot app is now generally available, so users can go from a chat to a Word draft or a PowerPoint deck and then hand off to the full desktop apps. Today, Pages and Workspaces aim to make Copilot more proactive and collaborative. Worth a look for anyone planning Copilot training, because where users start their work is shifting.
    https://techcommunity.microsoft.com/blog/microsoft-copilot-blog/office-at-ignite-2026-from-apps-to-agentic-system-of-work/4572226
  • Autopilot: meet your trusted personal teammate
    Autopilot is Microsoft’s always-on personal agent. It has its own identity, memory, workspace and cloud computer, works in Teams and Outlook, and can run routines without being asked. Still early (it’s only rolling out to eligible Frontier customers with a Copilot licence), but it’s the agent with its own identity that you’ll need to govern in client tenants.
    https://techcommunity.microsoft.com/blog/microsoft-copilot-blog/autopilot-meet-your-trusted-personal-teammate/4572611
  • The End of Starting Over: How Work IQ helps Copilot understand you and your work
    Work IQ is getting better memory and context, so Copilot remembers your preferences and picks up where you left off instead of being re-briefed every session. That should make Copilot more useful day to day. It also means clients should know what Copilot remembers and how they can manage it.
    https://techcommunity.microsoft.com/blog/microsoft-copilot-blog/the-end-of-starting-over-how-work-iq-helps-copilot-understand-you-and-your-work/4571169
  • Copilot that understands your organization: Introducing people and org context in Work IQ
    Work IQ adds governed people and org context, so Copilot and Cowork can answer questions about teams, structure and skills from authoritative org data. Microsoft admits the answers fall short when people data is poor. Good reminder to clean up Entra profile fields like manager, department and job title before clients expect this to work.
    https://techcommunity.microsoft.com/blog/microsoft-copilot-blog/copilot-that-understands-your-organization-introducing-people-and-org-context-in/4573356
  • MC1536065 – Microsoft 365 Copilot: Suggestions to continue in another Copilot experience
    After a chat prompt, Copilot will suggest switching to a better-suited mode such as Cowork and carry the conversation context across. It’s on the web, rolling out from mid-November 2026 for Standard and mid-December for Deferred audiences, with no admin action needed. A heads-up for helpdesk teams before users start asking why Copilot is suggesting a switch.
    https://mc.merill.net/message/MC1536065
  • MC1552389 – Microsoft 365 Copilot: Workforce Intelligence plugin available in Copilot Cowork
    A Workforce Intelligence plugin in Cowork lets leaders explore org structure and skills in plain language, respecting existing data permissions. Note the limited rollout: it’s only for tenants enrolled in Frontier, needs admin setup, and Microsoft says it isn’t for employee evaluation or performance management. Most SMB clients won’t see it yet.
    https://mc.merill.net/message/MC1552389
  • Rethinking Windows management for modern IT
    Windows Autopatch gets more control over update approvals, with quality update policies that let you choose automatic or manual approval for security, preview and out-of-band updates. Microsoft also points to hotpatch for no-restart security updates and to driver status in update decisions. Worth a look if you’re still patching client fleets by hand.
    https://techcommunity.microsoft.com/blog/windows-itpro-blog/rethinking-windows-management-for-modern-it/4572664
  • How to Update All Apps on a Windows Device with Winget
    My latest step-by-step guide to updating every app on a Windows device with winget. It covers checking winget is installed, listing and upgrading apps, silent switches, pinning apps you don’t want touched, and scheduling. A quick, free way for MSPs to close the third-party patching gap.
    https://blog.ciaops.com/2026/11/19/how-to-update-all-apps-on-a-windows-device-with-winget/
  • Microsoft Ignite 2026 Book of News
    Every key Ignite 2026 announcement in one interactive guide, grouped by topic. It’s a lot to get through, so use it as a reference rather than reading it end to end. Good for checking the detail on anything you hear about from vendors or clients this week.
    https://news.microsoft.com/events/ignite/2026/book-of-news/

Cloud & AI

  • Seven new world class models at scalable price-performance
    Microsoft AI released seven models at Ignite, including MAI Code 2, MAI Cyber 2 Flash, MAI Image 2.7, MAI Realtime 1 and Decision 1.1, all pitched on cost per task. Microsoft claims MAI Code 2 reaches 96% of Opus 5.5’s performance at 5% of the price. Those are vendor benchmarks, so test them yourself, but the cost of running agents all day keeps falling.
    https://microsoft.ai/news/microsoft-ignite-2026/
  • I ran MAI Image 2.7 through my usual espresso machine test
    My latest image model comparison, adding MAI Image 2.7 and MAI Image 2.7 Flash to my espresso machine test, with the results side by side. I also cover what Microsoft has actually published about the models and what that means for businesses already using Copilot. Useful if you’re choosing an image model for business work.
    https://blog.ciaops.com/2026/11/18/i-ran-mai-image-2-7-through-my-usual-espresso-machine-test/
  • GPT-6.2 Astra and GPT-6.2 Sol now generally available in Microsoft Foundry
    OpenAI’s GPT-6.2 Astra (reasoning) and GPT-6.2 Sol (production agents) are GA in Foundry. Microsoft says Astra gets better results with fewer output tokens at the same per-token price, so cost per task comes down. Good reminder to compare models on cost per task, not price per token.
    https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/gpt-6-2-astra-and-gpt-6-2-sol-now-generally-available-in-microsoft-foundry/4574122
  • GPT 6.2 Astra beats Sol and Opus 5.5 in my latest Cowork evaluation
    My latest Copilot Cowork model evaluation has GPT 6.2 Astra scoring 8.93, ahead of GPT 6.2 Sol on 8.78 and Opus 5.5 on 8.69. The gaps are small, and I explain what the test does and doesn’t show. Worth a look before you pick a default model for research-style work.
    https://blog.ciaops.com/2026/11/19/gpt-6-2-astra-beats-sol-and-opus-5-5-in-my-latest-cowork-evaluation/
  • Continuous improvement comes to Microsoft Foundry
    Foundry can now feed agent traces and user feedback back in, then find and test improvements and show the trade-offs in quality, speed and cost before you choose what ships. Options range from small skill tweaks to retraining model weights. Still early for most SMBs, but it’s the right model: agents that get better with use, with a person deciding what changes.
    https://azure.microsoft.com/en-us/blog/continuous-improvement-comes-to-microsoft-foundry/
  • Introducing Gemini 4 Neon
    Google’s Gemini 4 Neon is a fast frontier model aimed at software engineering, knowledge work and science tasks. It’s priced at $3 per million input tokens and $15 per million output tokens, with a 1 million token context window. Another strong option outside Microsoft’s stack, and pressure on everyone’s pricing.
    https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-neon/
  • Orchestrator: One API for reliable inference
    Perplexity’s Orchestrator puts open-weight and proprietary models behind one API that accepts OpenAI and Anthropic request formats. If a provider errors, rate-limits or times out, it retries on another deployment of the same model, such as Claude on AWS Bedrock. Worth a look if you’re building client automations and a single provider outage would break them.
    https://www.perplexity.ai/hub/blog/orchestrator-one-api-for-reliable-inference
  • Building an accounting team with Grok Bot
    SpaceXSI describes five Grok Bots that each own an accounting job end to end: provisioning, billing, billing support, cash applications and collections. A person only gets involved when something needs judgement or approval. It’s a practical model for MSPs: give each bot one well-scoped process, not a grab bag of tasks.
    https://x.ai/news/grok-bot-accounting
  • How to set up a Microsoft 365 help desk bot in Grok Bot, safely
    My latest guide to building a Microsoft 365 help desk bot in Grok Bot, with the safety rules first. It covers scope, connectors, Entra app registration, skills, routines, Teams, approvals and the extra care needed across multiple tenants. The bot takes the first pass, so your people spend their time on the problems that need judgement.
    https://blog.ciaops.com/2026/11/21/how-to-set-up-a-microsoft-365-help-desk-bot-in-grok-bot-safely/
  • How to let Grok Bot make outbound phone calls with Bland.ai
    Grok Bot’s built-in calling is US-only, so I connected a bot to Bland.ai to make outbound calls to Australian numbers. My post covers the connector, handling the API key safely, a first call, what comes back, costs, consent and troubleshooting. Worth a look if you’re outside the US and want a bot that can pick up the phone.
    https://blog.ciaops.com/2026/11/19/how-to-let-grok-bot-make-outbound-phone-calls-with-bland-ai/
  • How to set up voice in Grok Bot, and what is limited from Australia
    My latest guide to the voice features in Grok Bot: dictation, live voice chat, voice memos, phone calls and joining meetings. I’ve checked it against the official documentation and noted what’s actually limited from Australia. Spoken conversations and recorded calls need the same care as anything a bot sends in writing.
    https://blog.ciaops.com/2026/11/18/how-to-set-up-voice-in-grok-bot-and-what-is-limited-from-australia/
  • Your Grok Bot can have its own email address. Claim it carefully.
    Each Grok Bot account can claim one address at mail.grokbot.com, and plus aliases work too. My post covers how to turn it on, using it for sign-ups, and what to watch, including approval, privacy and MFA. A handy way to keep bot sign-ups out of your own inbox, if you set it up deliberately.
    https://blog.ciaops.com/2026/11/17/your-grok-bot-can-have-its-own-email-address-claim-it-carefully/
  • Grok Bot Privacy Mode: what it does, how to change it, and what that means for MSPs
    My latest look at Grok Bot Privacy Mode. It’s about model training and how model providers keep prompts, not a promise that Grok Bot keeps nothing. I cover what Cursor actually publishes, how to change the setting, and what it means if you’re working with customer Microsoft 365 data.
    https://blog.ciaops.com/2026/11/16/grok-bot-privacy-mode-what-it-does-and-how-to-change-it/
  • What MSPs should understand about Grok Bot security and customer Microsoft 365 data
    Many MSPs hesitate to point Grok Bot at a client tenant, and that’s reasonable. My post sticks to what Cursor and Microsoft actually publish about Grok Bot, Entra, Graph and partner access, and what I’ve set up in practice. If you can’t explain what a bot can reach and what happens to the data, don’t connect it to production yet.
    https://blog.ciaops.com/2026/11/15/what-msps-should-understand-about-grok-bot-security-and-customer-microsoft-365-data/
  • A Significant Increase in Digital Labor Automation
    This is from July, but it’s worth a read. The Remote Labor Index measures how often AI agents complete real paid freelance projects to a standard a client would accept, and Fable 5 reached 15.8%, roughly double Opus 4.8’s 8.3%. Still a small share of the work, but it’s climbing fast, and that’s where your clients’ process automation conversations are heading.
    https://safe.ai/blog/significant-increase-in-digital-labor-automation
  • Ignite 2026: Building the frontier ecosystem
    An edited version of Satya Nadella’s Ignite keynote, which casts Copilot as ‘the new OS for work’. He builds Microsoft’s stack around three principles: keeping human ambition at the centre, giving organisations control of their own intelligence and learning, and trust. Read it for the strategy behind the product announcements, not the detail.
    https://blogs.microsoft.com/blog/2026/11/17/ignite-2026-building-the-frontier-ecosystem/

Other

  • Annual currency adjustments to Microsoft Commercial Cloud pricing
    From 1 January 2027, Microsoft cloud prices in Australian dollars drop by about 4%, while NZD rises 10% and INR rises 15.7%. Individual SKUs may vary slightly because of rounding. Good news for Aussie MSPs and clients, so factor it into renewals and quotes that cross into January.
    https://news.microsoft.com/source/2026/11/10/annual-currency-adjustments-to-microsoft-commercial-cloud-pricing/
  • New partner benefits announced at Microsoft Ignite
    Microsoft has added more discounted AI licences to all partner benefits packages at no extra cost, and launched a new Partner Success Advanced package. Packages start at $350 a year for Partner Launch. Worth checking what’s now included, because it’s the cheapest way to run Copilot and AI internally as ‘Customer Zero’.
    https://partner.microsoft.com/en-US/partnership/partner-benefits-packages-benefits
  • Leading Frontier Transformation together at Microsoft Ignite 2026
    Nicole Dezen’s Ignite partner post covers new investment in the Microsoft AI Cloud Partner Program and pushes partners to adopt AI internally first. Microsoft’s study claims partners with high AI maturity earn 67% more partner revenue per dollar of Microsoft AI revenue. Mostly a direction-setting read, but it tells you where incentives are heading.
    https://partner.microsoft.com/en-US/blog/article/microsoft-ignite-2026
  • Anthropic starts trading on Wall Street
    Anthropic is now listed on the Nasdaq as ANT, opening at US$111.50 a share against an IPO price of US$83, which valued it at about US$1.5 trillion. Its S-1 shows US$16.3 billion of revenue in the first half of the year. Claude models sit inside Copilot, so a well-funded, public Anthropic is good for the vendors MSPs depend on.
    https://www.theverge.com/ai-artificial-intelligence/1035384/anthropic-ipo-trading-nasdaq

After hours

GTA 6’s Billion-Dollar HYpe Machine – https://www.youtube.com/watch?v=_19PHik8_Jg

Editorial

If you found this valuable, the I’d appreciate a ‘like’ or perhaps a donation at https://ko-fi.com/ciaops. This helps me know that people enjoy what I have created and provides resources to allow me to create more content. If you have any feedback or suggestions around this, I’m all ears. You can also find me via email director@ciaops.com and on X (Twitter) at https://www.twitter.com/directorcia.

If you want to be part of a dedicated Microsoft Cloud community with information and interactions daily, then consider becoming a CIAOPS Patron – www.ciaopspatron.com.

Watch out for the next CIA Brief next week

Leave a Reply