Your inbox isn't the problem. Neither are your meetings. It's the hours you spend on work AI could do in minutes. 

285 copy-paste prompts: built for email, meetings, admin, and delegation. Each one has a clear role, a specific task, and a defined output. 

No setup. No prompt engineering. Just copy, paste, and get your time back. Free when you subscribe to Mindstream.

Get The Free Guide

I'm writing this one from somewhere between a stack of boxes and a roll of parcel tape, because we're in the middle of moving house, but as they say, the show must go on.

After 12 years in the same place it's hard not to stop and take stock, and I'd got about as far as staring down the hallway, reminiscing about bringing my two boys home, England beating Colombia on penalties in the 2018 World Cup and the depths and despair of lockdowns, when my wife reminded me that we move in three days and handed me a box to get on with it.

So, having carved out a bit of time between the packing, here are three things from the week that just was: ChatGPT's rumoured new tool for businesses, which looks a lot more like Google Business Profile than the Search Console comparison doing the rounds, a new face at Obsero, and why some of the Muse tracking being talked up right now isn't quite what it seems.

The story doing the rounds this week is a screenshot from Juozas Kaziukėnas, founder of Marketplace Pulse (worth a follow), showing an unreleased "Claim your business in ChatGPT" page.

Barry Schwartz picked it up for Search Engine Roundtable and, like most of the coverage since, framed it as OpenAI building its own Search Console.

According to the screenshot, a verified business would be able to see how often it's mentioned in ChatGPT conversations, add its latest products and keep its business information up to date, and set up a business agent that answers customer questions with context the business controls.

Juozas says the analytics side counts impressions and clicks each day, and that it all sits inside a wider Business Hub that's only open to businesses that have verified ownership, for example by connecting a Shopify store.

I don't think the Search Console comparison holds up as well as the Google Business Profile one. Search Console is built around queries, so its whole value is seeing which searches you appeared for and how each one performed, whereas this shows no prompts, no user data and no competitor benchmarks, just a daily count of how often you appeared and how many people clicked.

Put that next to the rest of the page and it reads far more like Google Business Profile. You claim your business, verify you're allowed to represent it and keep your details and products current, with a simple view of impressions and clicks alongside. The business agent is the one genuinely new part, and it's essentially the AI-native version of a business answering customer questions directly on its own profile.

If you're a regular reader, this won't come as a huge surprise. Back in July, when I looked at how local businesses show up in AI search, the map in ChatGPT's local answers was being filled with Google Business Profile data, right down to the tracking parameters on the website link. Relying on a competitor's listings for something that commercially important was never going to last, so OpenAI building its own profile layer was always the logical next step.

Honestly, not a huge amount, because there's no launch date, no help page and no confirmed markets. There are a few things worth getting in order, though, so you're not starting from scratch when it does open.

First-party numbers from OpenAI would be a fantastic addition for marketers, and if OpenAI makes that data available, we'll bring it into Obsero.

Just don't expect OpenAI to tell you which prompts you're winning.

Obsero keeps growing, and this week we welcomed Kexin Yi to the team as our new founder's associate.

Kexin will be working closely with Mike and me across operations, commercial work and growth. As we build out the team she's taking on policy, compliance and the day-to-day running of the business, which frees me up to spend more of my time on product, and although she's only just started she's already spotted plenty of places where we can streamline how we work.

She's recently finished a master's in Strategy, Entrepreneurship and Innovation at King's College London, and before that she worked at Sodexo in Paris and ran her own travel business, where she looked after everything from marketing to operations. She's also worked in SEO and website marketing, so the way AI is changing how people discover brands is something she already finds fascinating.

What really stood out in the interview, though, were the questions she asked. They made it clear she believes in what we're building, and she has exactly the kind of resilience and grit you need at an early-stage startup. You'll probably see her at a few events in the coming months, so do say hello.

Here's my welcome post on LinkedIn if you'd like to connect.

Kexin won't be the last to join, either. We're hiring for four roles right now, a lead data scientist, a lead data engineer, a head of customer success and a GEO content specialist, all UK remote, so if that's you or someone you know, take a look at our careers page.

Meta's personal agent, Muse, has had a remarkable couple of months. Since launching in the US and Canada on 8 September it has passed 10 million users, with around a quarter using it daily, and Mark Zuckerberg told investors on Meta's Q3 call that shopping is now its most common use, with more than a quarter of users making purchases that average over $200. According to Sensor Tower, only 11 other apps have reached 10 million US downloads in their first 40 days.

Muse isn't the only one, either. Instinct, an invite-only personal agent you text or call, only launched in August and has already raised $1 billion at a $10 billion valuation, with its founder saying 35% of users rely on it for personal shopping. It hasn't shared user numbers (that I can find), but agents that know exactly who they're shopping for are clearly where this is heading.

So it's no surprise that a couple of AI visibility vendors have rushed out announcements saying they now track Muse. Read the small print, though, and what they actually track is Muse Spark, the model Meta launched in April that sits underneath Muse and that developers can call through Meta's API.

When one of them announced it last week, I asked the question in the comments: Muse needs a login, so are you running prompts through logged-in accounts or through the Muse Spark API? It's a simple question, and the answer tells you exactly what you're buying.

Muse isn't just the model. It's the model plus everything Meta knows about the person using it, their connected accounts and a set of workflows that let it act on their behalf, so it knows exactly who it's shopping for in a way the model on its own never can. Some vendors are being far more transparent about that difference than others.

That's the gap. A prompt sent to the model through the API carries no memory, no personal context and no connected accounts, so it can tell you what the model knows about your brand, but not whether Muse will recommend you to a particular shopper. That's still useful, as long as everyone is clear that it's model visibility and not what a customer actually sees.

I'd rather be clear about this stuff than let a dashboard imply something it isn't. Obsero tracks logged-out environments, which is what a customer sees when they use a platform without signing in, and where a platform only works behind a login we use the model instead and say so. Claude is the obvious example, where we use the API rather than the Claude app, because tracking a logged-in environment raises privacy and legal issues we're not willing to gloss over.

We'll most likely add Muse Spark once it's available globally, and when we do, it'll be labelled as the model. If a vendor tells you they track Muse, ask them one question: is that the agent your customers use, or the model underneath it?

That's it for this week. Until next time. 🏠

Have a good one.

Andy