Most people assume AI shopping recommendations fail because the picks are bad. Instinct’s rollout shows something else. The picks can be fine and users will still hate them if they didn’t ask for them.
According to TechCrunch AI, the AI agent startup Instinct has launched Instinct Selections, a feature that sends users curated product and place recommendations for dining, travel, and shopping. It went live overnight, and some users were pushing back within hours.
What Instinct actually launched
Founder Noah Shinn announced the feature on X. He said Instinct is partnering with “local chefs, designers, architects, travel guides” and others to bring “human taste” into the agent. So instead of relying only on what the model learned from the open web, Instinct can now draw on lists picked by real people.
Shinn gave three examples:
- Dining: restaurant lists “handpicked by chefs who know the hidden gems in your area”
- Outdoors: trail routes suggested by local guides and matched to your preferences
- Home: pieces from independent designers that fit “your style, space, and budget”
“Our goal is to deliver world-class recommendations that are differentiated in quality from what you might find on the internet, and from what other chatbots produce,” Shinn said.
Instinct hasn’t said who the human curators are. It also hasn’t said whether it makes money from these recommendations or plans to. TechCrunch AI notes that the feature looks ripe for an advertising, commerce, or affiliate play.
The backlash is about consent, not quality
If the complaints were about bad picks, better curation would fix them. They aren’t. People are objecting to getting pitches they never asked for.
- Shruti Gandhi, general partner at Array VC: “I didn’t need to buy carry-on luggage, sunglasses, or a brim hat. This feels so disappointing. At least make me a personalized shopping list of things I actually need!!”
- Andrew Yeung, entrepreneur, said on X that he also got product suggestions he hadn’t requested.
- Chat Joglekar, founder of a business-buying marketplace: “I had my first ‘ewwww’ moment with an AI assistant when Instinct started to suggest products that I should buy based on what I had in email and upcoming trips.”
That last quote is the one Instinct should worry about most. Joglekar isn’t saying the picks were bad. He’s saying the agent took what it knew from his inbox and travel plans and used it to sell to him. That’s what turns “helpful” into “creepy.”
Why this is a bigger risk for an agent than a chatbot
An AI agent only works if you give it deep access: your email, your calendar, your payment details. You hand that over because you trust it to act for you. Once it starts sending you unsolicited shopping suggestions, you start wondering whose side it’s on.
TechCrunch AI puts the fix plainly. Recommendations need to answer something the user asked for, or at least fit naturally into the conversation. If they show up at random, they’d better be excellent and highly personal, or they’ll feel like spam.
The timing doesn’t help either. Instinct is competing with agents from Meta and now OpenAI, and according to the report, those agents don’t do this. When the big players feel cleaner, users have an easy place to go.
The money behind the move
The launch came shortly after Instinct confirmed a $1 billion Series C at a $10 billion valuation. The company hasn’t shared how many people use the product, and that number would go a long way toward justifying the price tag.
Shinn did offer one data point. In a podcast with investor Patrick O’Shaughnessy, he said the platform is approaching $1 billion in annual transactions handled by the agent, and more than half of them are travel-related.
That explains the strategy. If your agent already books trips, recommending luggage and sunglasses for those trips is an obvious revenue stream. But being obvious doesn’t make it welcome.
What to watch
Instinct is testing a question the whole agent industry will have to answer soon: can an assistant that acts for you also sell to you? Getting the timing right, being open about any paid partnerships, and adding an easy opt-out would go a long way. Whether Instinct makes those changes will show how much it values user trust compared with a new revenue line.
The full story, including Instinct’s pending response, is available at TechCrunch AI.