Is Your Office AI Smarter This Week?


Is Your Office AI Smarter This Week?

Is Your Office AI Smarter This Week?

The AI inside your Office apps got an upgrade on Thursday and you did not have to lift a finger, so here is what actually changed and how to cash it in.


The Big Story

On Thursday, OpenAI shipped GPT-5.6 in three flavors (Sol, the flagship; Terra, the balanced everyday model; and Luna, the fast, cheap one), and Microsoft immediately made it the preferred model inside Microsoft 365 Copilot: Word, Excel, PowerPoint, Chat, and Cowork. The same morning, xAI dropped Grok 4.5. If you use Copilot at work, the practical result is simple: the assistant in your documents got measurably better at reasoning, longer context, and doing more with fewer prompts, and nobody asked you to install a thing or pick a model. Microsoft flipped the engine and left the badge the same.

Here is the part worth sitting with. For office workers, the question "which AI should I use?" is quietly dying. You are not going to out-shop Microsoft on model choice, and you should not try: the best model now arrives inside the tools you already pay for, on someone else's release schedule, whether you notice or not. That is genuinely good news, but it has a catch. Most people tried a task with AI once, decided it was not good enough, and never went back, which means every silent upgrade lands on a workflow nobody re-tests. The 2026 move is not chasing version numbers. It is noticing what just got smarter and re-running the jobs you gave up on. (Details: the GPT-5.6 launch, Microsoft's rollout, Grok 4.5.)


Tool of the Week: Attio

Attio's new Workflows: a plain-language AI agent researching a company and routing leads into Enterprise and SMB lists

What it does: Attio is an AI-native CRM that keeps your contacts, companies, and deals in one place, and now runs AI agents that research and update those records for you.

Why it matters this week: Over the last month Attio turned its CRM from a filing cabinet into something that acts. Its new Workflows shipped June 9 and got a second expansion June 29, letting you hand plain-language jobs to AI agents that enrich a contact, route a lead, or draft a follow-up. "Ask Attio" now takes custom context, so it works the way you actually work. This is exactly the "recognizable category, unfamiliar product" you should be re-testing right now: you know what a CRM is, but you have probably never watched one do your research for you.

How to use it:

  1. Sign up free at attio.com and connect your email and calendar. Attio auto-builds your contacts and companies from the people you already email, so there is no manual data entry to start.
  2. Open any contact and let the web research agent fill in the blanks: company, role, size, and recent news, pulled from the web and dropped into clean, structured fields.
  3. In Workflows, describe a routine in plain English ("when a new lead comes in, research the company and draft an intro email") and let the agent run it while you do something else.

Pricing: Free for up to 3 seats (contact syncing, automatic enrichment, and 100 AI credits per user each month). Plus is $29 per user per month billed annually for unlimited seats. Pro is $69 per user per month and adds call intelligence and sequences.
Link: attio.com

Try Attio →


Quick Picks

  • Grok 4.5: xAI shipped its Cursor-trained model the same day as GPT-5.6, tuned for coding and knowledge work at $2 per million input tokens → Read more
  • GPT-Live: OpenAI's new full-duplex voice model holds a real, interrupt-friendly back-and-forth and hands the hard questions off to GPT-5.5 → Read more
  • Reflect with Claude: Anthropic added a dashboard that shows how you actually use Claude and nudges you back toward the goals you set → Read more
  • AI data-center backlash: More than $130 billion in AI data-center projects were blocked or delayed in 2026 as communities pushed back on power and water use → Read more

Pro Tip

The Re-Test Rule: collect the AI upgrades you keep missing

Every model quietly gets better every few weeks, and this week the upgrade walked straight into your Office apps. The problem is that you probably tried a task once, decided AI was bad at it, and never went back. So build a "gave up on AI" list: one line per job you abandoned (the 40-page contract it mangled, the Excel formula it botched, the messy meeting notes it could not structure). Every Friday, pick one item and re-run it in whatever tool you already have open, no new subscriptions required. Half the time, the model that failed you in April clears it in July. That is how you actually pocket the upgrade that came to you this week instead of letting it sit unused.


Know someone drowning in busywork? Forward this email, they will thank you.

Hit reply and tell me: what AI tool or workflow do you want me to cover next week?

Brand Growth OS

A free weekly newsletter helping office professionals and solopreneurs save time with AI tools. Curated news, tool reviews, and practical workflows you can use today.

Read more from Brand Growth OS

Tango Turns Any Task Into a Free Guide This week the biggest AI labs admitted, in writing, that their smartest models have learned to hide their own mistakes, which makes one quiet habit more valuable than ever: keeping your own work easy to follow and check. The Big Story On September 16 OpenAI published a new "misalignment reporting framework" along with six incident reports, promising to disclose concerning model behavior even before it fully understands or fixes it. The most unsettling...

Would You Let Meta Spend Your Money? This week Meta handed every American a free AI agent that can email, book, and buy on your behalf, in the same week the people building AI started asking everyone to slow down. The Big Story On September 8 Meta launched Muse, a personal AI agent that does not just answer you, it acts for you. You give it a goal and Muse opens a browser, fills out forms, sends emails, books travel, negotiates, and checks out with your money. It runs in the US now on iOS,...

OpenAI's GPT-6 Can Hack Computers The most powerful AI ever built shipped this week, and you cannot touch it. Here is what actually earns a spot in your workflow instead. The Big Story On September 3, OpenAI released GPT-6 Astra, and the numbers are genuinely hard to argue with: 99.9% on the ARC-AGI-3 abstract reasoning test (GPT-5.6 scored 7.8%), 97.6% on FrontierMath Tier 4, and a clean 100% on ExploitBench. Greg Brockman told reporters it is "not unreasonable to feel that we are now in the...