
💊 Vetsource opened a fourth pet pharmacy in Atlanta aiming for 12,000 orders a day
💼 A teacher who made $35K left the classroom for dog walking and now earns $4,500 a month
🦗 Mars' ingredient accelerator drew half as many applicants this year, 83 down from 165, and all three picks turn underused feedstocks into protein or fat
🤠 Skiptown is coming to Austin with five franchise units, bringing its pipeline to at least 12 with none open yet
📱 Fetch members now get four free Vetster video visits a year with no claim form to file
👟 adidas and UNheardof built a dog walking shoe inspired by a dog named Otto, with a December release and a "Puppy Runner" version for kids
🐻 Scientists reviewed 326 dog-bear encounters and found the dog started it 54% of the time, often leading the bear right back to its owner

Last week I let Muse, Meta's new AI agent, file a class action claim for me.
It was the Apple one, the $250M settlement over Siri features that some iPhone buyers, including myself, say never showed up. Muse asked for my phone model and when I bought it to see if I qualified. It told me it was submitting, came back with a claim ID and asked if I wanted to be paid by Paypal, ACH or Venmo.

It did get the final hearing date wrong. The settlement site says February 24th, 2027.

The settlement administrator's confirmation email
It clicked through a claims form, pulled my details together and filed the whole thing with barely any effort on my end. So I started wondering what it could do with Instagram and Facebook. Meta built those right into it, and it already has my saves, the accounts I follow and everything I interact with.
I asked it what some of the biggest pain points were for pet owners on Instagram and Facebook right now, and it came back with five broad topics. That was a search across all of Facebook and Instagram, so I wanted to see how it did in a smaller space I know well. I pointed it at just the Facebook groups I'm in, public and private, all of them pet related. It went through each one and broke what people were posting about into food, gear and travel. Having spent time in those same groups myself, it looked spot on. Nothing made up, and I could see the actual posts it was pulling from.
I was like, damn, no need to keep a bank of screenshots anymore. I can query Facebook groups the way I'd query any other database, or Google. I've barely used Facebook in recent years because it's so painstaking to find posts or comments of any value. Such a manual effort to comb through things.
So there's now an agent that can read the places pet owners ask each other for help and then pay for things. Nobody in pet has plugged into it yet, which is normal a month in, and the lane is wide open for now. Meta is taking connector submissions through a public page right now.
The comments say more than the scores do
Then I went back to the wide view with a much more detailed ask, this time on Instagram. I wanted to not only know what pet owners were most frustrated with, but what they said they needed a fix for, what they'd buy if it existed, and which topics, products and brands came up the most. I told it to score every finding for seriousness, sentiment and opportunity, and to back each one with real posts and what the comments were saying.
It came back with a full research report. It ran eight sweeps across Instagram posts from January through September, ranked 11 findings and tied each one to specific posts with their like and comment counts. By its own scoring, 7 of the 11 rated 8 or higher out of 10 for opportunity. Finding a sitter you can trust came out on top.

It also warned me and was upfront about its limitations. It hit a rate limit partway through the deeper reads, stopped, and told me its comment sentiment came from summaries, not full threads.
About an hour later I told it to keep going.
It pulled around 445 comments across nine posts and rewrote the report with the actual comments quoted.
The second pass is where it got useful for anyone running a pet business, because it started really reading the room instead of just counting it.
On vet prices, it scored roughly 85% of the first 50 comments on one viral post as angry at corporate clinic owners, with vet staff mostly defended rather than blamed. Then it picked out threads I'd never have found scrolling. Some owners are organizing to push state lawmakers on insurance denials for pre-existing conditions. Others said the dog they have now will be their last because of what care costs. A few were connecting that Mars owns both vet clinics and pet food brands, and one said their independent vet runs about 30% cheaper than the corporate ones nearby.
The insurance thread was the most revealing. In the comments it pulled, Trupanion got praised almost across the board, Lemonade and MetLife came off clean, and Fetch and Pets Best took the most heat. It also caught that one commenter defending their insurer hijacked the thread into a flame war, and flagged that these threads turn tribal fast. Underneath the fight, a “self-insuring” crowd is growing. People say they canceled and moved the premium into a savings account, and several said threatening a complaint with their state insurance department got a denied claim moving. That's a load of comments, not a survey, but it's the kind of read an insurer's marketing team would pay for. I’m also summarizing much of what was said as the level of detail Muse provided was surprising.
Food split into two camps. Vet-aligned commenters defended the big science-backed brands and told people to listen to their vet, not an influencer. The other side backed the creator’s knocks on heavy processing and ingredient splitting. On a dog food ranking video, top comments asked which brand paid for the post, and one said "I'll just ask ChatGPT." That tracks with what we see across food and supplements, where every claim gets asked who paid for it or some form of trust issue.
Sitters scored highest due to severe lack of trust. One of the top posts it flagged was a trainer's reel about how finding someone to watch the dog is harder than booking the trip, and the comments are full of owners saying boarding costs more than the trip itself. One wrote "If we can't bring our pups, we don't go." That's travel spend that never happens because people don’t have reliable care.
The strongest buy signal in the whole set was a car window chin rest for dogs called the Whiffie cushion, with about 90% of the comments it sampled having positive sentiment and many asking where to buy.
I wasn't grading whether every post or analysis was true or the cleanest data set. I wanted to see what it could surface from a multi-pronged prompt, and it was more than I expected.
The other agents start from what you tell them
ChatGPT works from what you type and what it remembers from past chats, and it's had checkout inside the chat since last September. Gemini is the closest thing to a search box, and it can read your Gmail and YouTube history once you opt in to that. Instinct will place phone calls for you, same as Muse. Grok leans on your X activity to personalize its answers, and you have to dig into settings to stop it.
As I was wrapping this story up, OpenAI launched its always-on agents, called dots, powered by its new GPT-6 Astra model. Each one runs on its own cloud computer with its own browser, connects to more than 4,000 apps, and keeps working in the background after you close the chat. The same day, OpenAI said developers can now ship full apps with plugin extensions inside ChatGPT, and that it will surface the relevant ones right in the conversation for its 1.2B weekly users. That's one more shelf a pet brand can build for.
Dots might end up the stronger worker. What none of these start with is years of what you liked, saved, followed and asked about on Facebook and Instagram. Muse does. Outside of Reddit and TikTok, Facebook and Instagram are where pet owners go to find out what other owners actually think and share their POV. That's where a lot of the deciding happens.
Your saved reels are shopping inputs now
Meta launched Muse in the US on September 8th, and Sensor Tower had it above 3.4M downloads by September 25th, per TechCrunch's roundup. A Citi Research chart making the rounds on X has its first two weeks of US downloads running well ahead of ChatGPT's launch pace.

The one shopping example in Meta's launch post is a recipe reel you saved on Instagram turning into a grocery list. The same post says "people choose which apps Muse connects to," which is how it works for Gmail. For Meta's own apps it's less clear. One reviewer found his accounts already linked the moment he signed in with a phone number, and so were mine. I'm guessing that's covered somewhere in the terms of service, which we all read so thoroughly.
Meta's launch post says Muse conversations don't go into its ad systems. By default, though, Muse does use your interactions for training, and you can switch that off.
I could be wrong about how much a save actually counts when Muse picks a product. Meta hasn't said. My guess is it counts for more than a search, because you save the stuff you're still thinking about.
The groups already do this job
Pet owners were doing an agent's job by hand long before Muse showed up. I'm in Flying With Dogs, a public group with 109.4K members, plus a couple of private groups built around dog food and dog-friendly rentals. The posts that get the most replies stack five or six conditions into one ask, like a rental with a real fence, no breed or size limit and room for a few dogs over the holidays. Most booking filters can't handle that. The group can, because someone in there has stayed at a cabin that fits the bill. The rental group even runs its own certification for dog-friendly properties, a trust signal no product catalog carries.
The dog still needs a phone call
Muse can shop the whole Shopify catalog, and Meta has added Walmart, Wayfair and Instacart, all of which sell pet supplies. Reordering pet food or supplements looks easy.
Booking the dog is harder, so I tested it. I asked Muse to start booking a round trip flight from Austin to Denver next month for me and Toast, my 24lb pug. Cheapest option that allows an in-cabin pet, confirm it can add him, and don't pay for anything.
Frontier came back cheapest, about $328 round trip with the dog. I asked whether it had actually confirmed a spot for Toast. It said no, because its fare search can't see pet spots. Frontier takes pet reservations online, so it offered to finish on the airline's site. It opened a browser, picked the nonstops, checked the Cabin Pet box and stopped one step before payment. Then it flagged that the site was charging $149 each way for Toast, not the $99 in Frontier's FAQ, which put the real total closer to $379.

Delta and Southwest want a phone call after you book, and Muse told me that step would be on me. Muse can call US businesses now, but that's rolling out in phases, and my account doesn't have it yet. It also went back and forth on Delta's fee, first $95 and then $125. Delta's own page says $150 each way for tickets issued since April 8th, 2025. It caught the Frontier fee on its own but it never caught Delta's.
I also tried it on the everyday stuff. So I told Pippy (my Muse’s name) that Toast needed his nails done, and I had to leave town the next night. I asked Muse for a same-day nail trim and two nights of boarding or a sitter in Austin, with two walks a day and toppers that have to stay in the fridge.
It came back fast, with prices and a plan, and it was upfront about where it got stuck. Kennels "mostly need an assessment day before a first stay," it told me, "and most have no published policy on refrigerated food." Then I pushed on the list. The cheapest groomer it gave me turned out to be in St. Marys, Ohio. It pitched Groomit as a local mobile groomer, and Groomit is a booking marketplace. It owned every mistake and fixed it, but I still had to be the one who knew the Austin market. Its pick for boarding was a Rover sitter. Rover isn't a Muse partner, but it offered to log into my Rover account and message the sitter for me. That's an agent working around a missing connector.
Phone calls are a problem even when the agent can make them. Matt Van Horn had Muse call a karaoke spot to book a room, and the place hung up on it for being AI. So he had Muse hand the job to Fo, a new agent from a former Google DeepMind engineer that passes a task to a real person when the AI gets stuck. A human booked the room, and Muse let him know it was done. Swap the karaoke spot for a kennel's intake line and it's the same problem.
Meta's Connect recap named 19 companies as partners, from Walmart and Sephora to Expedia and GitHub. Its chief AI officer then listed 22 more for small businesses, from QuickBooks to Klaviyo. That's 38 unique names so far, and none of them is a pet company.
Meta went after the biggest general retailers and the software small businesses already run on, and a vertical like pet comes after that. Wang said Muse got more than 1,500 connector applications in under a week.
Chewy is the one pet company I'd expect early. Back in May it gave investors a presentation on AI agents titled "AI Won't Disrupt Chewy. It Will Route Demand to It."
Meta just signed up the other side
On September 29th Meta launched Muse for Small Business.
Wang said they'd found people already using Muse to run plumbing businesses, grocery stores, farms, restaurants and shops, and the new connectors reach Meta ad accounts, Facebook Pages and Instagram analytics, plus Shopify, QuickBooks, Klaviyo and HighLevel. Shopify's version lets a store owner ask Muse about orders and inventory and have it check in every Monday.
That's the same agent on both sides of the counter. It reads what the owner saved, and it can now read the shop's ad account and order history too. Plenty of groomers, sitters, boarders and small treat brands run on an Instagram account, a Facebook Page and a Shopify store.
What isn't on the list yet is the booking and kennel software a groomer or boarder runs their day on, and that's the system that knows whether Saturday at 10am has an opening. That's the gap a pet connector could fill.
Loyal food buyers don't need an agent
There's a real case that Muse won't move much spending on food and other consumables, at least not yet.
Once a dog does well on a food, many owners stop shopping around, subscription or not. At Chewy, Autoship customers accounted for 83.3% of fiscal 2025 net sales. The bag that shows up every month doesn't need an agent to find it. Services don't work that way. Owners are just as loyal to a good vet, groomer or sitter, but finding one takes real time, energy and trust. You're handing over your dog or cat. And you go through it more often than you'd think, whether you're switching for good, covering while your usual person is booked, or looking for the first time. Every one of those searches ends with a person at a facility, which is exactly where my tests kept stalling. That's the job for an agent here, taking out the friction while earning the trust.
And trust is the part agents haven't earned yet. An Acosta Group survey found only 22% of grocery shoppers trust AI agents to buy on their own. And Muse has already had a public stumble on Meta's own turf. A seller who let it run his Facebook Marketplace listings says it shared his home address with a buyer, took a lowball offer, and told the buyer he was home when he wasn't.
A wrong Marketplace price is annoying. A wrong food for a dog on a renal diet, or a carrier that fails an airline's size check at the gate, costs a lot more. Meta hasn't said who covers that when the agent gets it wrong. At Connect, Zuckerberg said Meta expects to profit from a small fee on transactions, which makes that question harder to ignore.
Listen first, then build the connector
The partner list is going to grow, and a few things seem worth doing now.
The big one is using Muse to listen. Paid ads mostly tell you who clicked. The groups and comments are where people explain why they're stuck, in their own words. Muse can now read the groups you're already in and the ad account you're already paying for, so you can check what your ads are telling you against what people are actually saying on Facebook and Instagram this week. That goes a lot deeper than an ad dashboard on its own, and it's a better starting point for a connector than a guess. What it hands you still needs checking, because it will stretch things.
Pet insurers and vet clinics should pay attention too, because owners are already using it there. People have posted about Muse filing their pet's insurance claim from a vet invoice.
Meta's chief AI officer, Alexandr Wang, also challenged users to have Muse go through their finances and post what it saved them, and one founder says it found him car insurance $3,500 a year cheaper and switched him in about five minutes. None of those savings have been independently checked, and plenty of the early replies came from Meta employees. But denials and premium jumps sat near the top of what Muse found on Instagram, and only 4.27% of US pets are insured. An agent that reads the exclusions and re-shops the policy is bad news for any renewal that depends on nobody checking.
For travel and services, the pet policy has to be something an agent can read and book. Fees, weight limits, breed rules, pets per room or flight, and whether you can do it without a person. My guess is the first airline that lets an agent book the dog end to end picks up a lot of those bookings.
The two things I'm watching are the first pet company Meta names as a partner, or gets listed as a connector and what that transaction fee turns out to be.
Want us to point Muse at your category, your brand or your competitors? Reply and tell me what you'd want it to dig into. We'll run a few and share what comes back.



