Every Friday I clear out a folder I only half-jokingly call “AI told me it changed again.” Product emails, changelog links, breathless posts about the future of work. Most of it does not matter to the person who just wants to write a cleaner email, build a deck, or get through a meeting. A small slice actually does.
This is that slice. Six updates from the last week or so across Copilot, Claude, ChatGPT, Gemini, and Zoom, ranked by how much they help a normal business user. No back-end settings, no admin console tours, no hype. If it will not change how you actually work, it is not here.
The ones worth your time this week
1. You can now pick Claude inside Microsoft 365 Copilot
Microsoft’s July Copilot release added Anthropic’s Claude as a model you can choose right inside Copilot Chat, next to Microsoft’s own models. You pick Claude for heavier analysis, document understanding, or structured writing, then switch back for everyday tasks. It is a real choice, not a label.
Why this matters to average users: if you already work in Word, Outlook, and Teams, you no longer have to leave for a second tool when a job needs a different kind of thinking. Source: Braintree, Microsoft 365 Copilot July 2026 updates.
2. Copilot Notebooks can turn your notes into a Word doc, Excel sheet, or slide deck
Copilot Notebooks used to be a place to collect notes. Now it can take that pile of half-formed thinking and build an actual first draft: a Word document, an Excel sheet, or a PowerPoint deck. It will also draw a mind map of what is inside so you can see the shape of your own notes.
Why this matters to average users: the blank page is where most work stalls, and this turns your messy notes into a rough draft in the right file type. Source: A Guide to Cloud, Microsoft 365 Copilot July 2026.
3. Claude’s voice mode got smarter
Anthropic updated Claude’s voice mode so it is no longer stuck on the quick, lightweight model. You can now use Opus, Sonnet, or Haiku by voice, and by default it picks up the model from your last text chat and runs its fastest version. Talking through a complex question now gets a capable answer instead of a rushed one.
Why this matters to average users: voice stops being just for setting timers and simple questions, and becomes useful for working through real problems out loud. Source: TechCrunch, July 23, 2026.
4. ChatGPT Voice got more natural and moved onto the desktop
OpenAI rolled out a more natural, real-time ChatGPT Voice that handles interruptions and follow-up questions, and can pull in a web search mid-conversation. On the desktop app it now goes further, letting you talk through work and check on tasks running in the background without typing.
Why this matters to average users: a back-and-forth voice conversation is closer to talking to a colleague than dictating to a machine, which is easier when your hands are busy. Source: OpenAI ChatGPT changelog.
Voice modes are the fastest way to test whether a tool actually understands your work. Ask it to explain something you already know well. If the spoken answer holds up, you can trust it on the things you do not.
5. Gemini Omni brings easier video into the Gemini app and Google Vids
Google put its Omni video model directly inside the Gemini app and Google Vids. You can generate short video from a description, and just as usefully, make plain-language edits to video you already have: fix the color, change the style, even ask it to remove a background noise like a passing siren. No editing software and no timeline to learn.
Why this matters to average users: short video for a training clip, an announcement, or a social post is now something you can describe in a sentence instead of hiring out. Source: Gemini Apps release notes.
6. ChatGPT will now remember a lot more about you
OpenAI raised the limit on ChatGPT’s custom instructions from 1,500 characters to 5,000. Custom instructions are the standing brief that shapes every answer: who you are, how you like things written, what context to assume. More room means you can spell out your role, your projects, and your preferences once instead of re-explaining them in every chat.
Why this matters to average users: a fuller brief means fewer generic answers and less time correcting the same things over and over. Source: reconnAI LLM changelog.
Still worth catching up on
A couple of items landed just outside the last seven days but are rolling out now and belong on your radar.
Gemini in Chrome keeps expanding. Google widened the rollout of its browser side panel, which summarizes the page you are on, compares information across open tabs, and connects to Gmail, Calendar, and Maps without leaving the tab. If it has not shown up in your Chrome yet, it likely will soon. Source: Google, expanding Gemini in Chrome.
Gemini writing help in Google Docs added 11 more languages. If your team drafts in more than English, the Docs writing and formatting help now reaches further. Source: Google release notes, July 2026.
The quiet one
Worth an honest note: Zoom did not ship a headline AI feature for everyday users this week. The recent changes are smaller housekeeping items, like being able to resend a meeting summary and add or remove recipients after the fact, and post-meeting emails now routing everything into one place. Useful if you live in Zoom, but nothing that changes your day. When it does, it will show up here.
A slow news week for one tool is not a problem. The pace of these updates is not a scoreboard, and you do not have to adopt something just because it shipped. The goal is fewer, better changes to how you work.
What to actually do with this
You do not need to touch all six. Pick the one that lands closest to a task you already dread. If you build decks, try Copilot Notebooks on your next set of notes. If you re-explain yourself to ChatGPT constantly, spend ten minutes on your custom instructions. If you think out loud, give a voice mode a real problem and see if it keeps up.
The pattern across every one of these is the same. The tools are moving from “answer my question” toward “help me finish the work.” That shift rewards people who try one thing on real work over people who read every changelog and change nothing.
Most AI news is noise. This week, the signal is that your everyday tools are quietly getting better at finishing the work, not just describing it. Pick one update, test it on something real, and skip the rest.
Jay Everson – IT Leader & Strategic Innovator with over 25 years in IT, Jay Everson is a highly accomplished professional known for his expertise in IT infrastructure, systems administration, and project management. Jay has led global teams, driven enterprise-level projects, and executed complex acquisitions and integrations. His leadership blends technical proficiency with a hands-on approach, guiding cross-functional teams to success. Recognized for resourcefulness, adaptability, and strategic vision, Jay is a trusted leader in the IT community.
