More than 1 in 3 Americans bought something based on an AI recommendation in the past year
44% say AI helps them find better value for money
34% would pay more for a product if AI showed it was the best option for them
Charts of the Week: a16z.news/p/this-week-in…
44% say AI helps them find better value for money
34% would pay more for a product if AI showed it was the best option for them
Charts of the Week: a16z.news/p/this-week-in…
Oct 16, 2026 · 8:45 PM UTC
61
59
413
65,010
@a16z The 34% figure is fascinating, but what exactly does "AI showed it was the best option" mean?
Did consumers independently validate the recommendation, or trust the AI's ranking?
Did consumers independently validate the recommendation, or trust the AI's ranking?
0
0
0
68
@a16z 1 in 3 americans let an AI pick their purchases and the other 2 are lying. best salesman ever built and it doesn't even want commission
1
0
2
318
@a16z if 1 in 3 already buy off an ai rec, fake reviews and fake shops are gonna aim at the ai next, not the shopper
0
0
1
170
@a16z 1 in 3 already let AI pick what to buy. If you still spend hours reading reviews for a $20 thing, you're probably a 废物.
0
0
0
5
@a16z 34% would pay more because ai said so is the wildest stat here
0
0
0
96
@a16z the 34% willing to pay more based on a rec is the real goldmine brands are about to weaponize
0
0
0
9
@a16z If 34% will pay more just because an AI picked it, retailers are going to fight hard for that slot.
0
0
2
167
@a16z 34% would pay more if AI says it's the best option. of course — the perfect salesperson is one that can show its work. the upsell just got a PhD
0
0
1
46
@a16z Would that 34% hold up once AI answers start carrying ads?
0
0
1
61
@a16z People already trust the AI's pick enough to pay more for it, but retailers are blocking the agents that would finish the purchase. Something could be said about adding friction to the process when there.
0
0
1
36
@a16z 1 in 3 people outsource their taste to a model and the other 2 just don't admit it yet.
0
0
0
6
@a16z Curious how this plays out for bigger purchases like travel…once people trust an agent to pick and book the whole trip end to end that's game changing
0
0
0
36
@a16z 34% paying more on an ai's say-so is the number i'd look at before the 1 in 3
0
0
1
31
@a16z A third of buyers following AI recommendations means 'rank well with the model' is the new SEO. Somebody is about to sell a very confident consultancy deck.
0
0
0
50
@a16z Banger of a chart. AI recommending products: fine. AI holding my credit card: absolutely not
0
0
0
20
@a16z This is actually impressive I think more of this is going to happen
0
0
0
8
@a16z Fascinating stats AI driving real purchase decisions is incredible
0
0
0
17
@a16z Read that again: more than 1 in 3 people bought something because an AI told them to.
If your product pages are thin, vague or missing real specs, the AI has nothing to recommend. You are invisible to those buyers and you do not even know it.
If your product pages are thin, vague or missing real specs, the AI has nothing to recommend. You are invisible to those buyers and you do not even know it.
0
0
1
27
@a16z 41% would consider buying a product they'd never heard of because AI recommended it. Good news for small brands that never won on brand search.
0
0
0
29
@a16z That 34% is the stat to sit with. A third of shoppers would pay more because a model said it's the best fit, so the new shelf space is the answer box. The honest catch: it only lasts while those recommendations stay earned, not bought.
0
0
0
3
@a16z4% value-for-money line is the one owners can act on this month. An assistant cannot compare a price it cannot see, so "contact us for pricing" pages and vague packages are easy to leave out of the answer. A clear starting price, what is included, and who the offer is not for give the model something concrete to recommend.
0
0
0
19
@a16z more than 1 in 3 Americans bought something on an AI recommendation in the past year, and 34% would pay more if AI showed a product was the best option for them. picture someone at checkout with the pricey kettle, telling the cashier "my assistant vouched for it."
0
0
1
70
@a16z 34% would pay more if AI said it's the best option, that's a crazy number for pricing pages
are brands even checking what AI says about them yet
are brands even checking what AI says about them yet
0
0
0
3
@a16z 34% would pay MORE if AI tells them it's the best option is the wildest stat here. we're really all just shopping by algorithm now lol
0
0
0
31
@a16z 三分之一的人让 AI 帮自己做购买决策，说明信任已经过临界点了。但 34% 愿意多付钱我存疑，问卷里的表态和结账时掏钱是两回事，价格敏感度在付款那一刻会翻倍。不过 AI 推荐+比价确实是离钱最近的一条变现路径，这点没毛病。
0
0
0
68
@a16z If a third of people already buy on an AI's recommendation, the store page matters less than what the assistant can read and compare. Fewer screens, more just asking. Where I think that leaves interfaces: x.com/uctive/status/…
Do we really need a UI for everything?
Based on my latest tests of new products with AI agents, I’m increasingly thinking that, in many cases, we simply don’t need a user interface at all.
Learning buttons, menus, and the logic behind each interface takes time. Every product nd even when developers do their best to make everything intuitive and beautiful, there’s still a learning curve. Products are also changing so quickly that by the time you get used to one interface, it may already have changed.
So I’m finding myself looking at UIs less and less.
Instead, I ask an agent to install the software and perform whatever functions I need — through an API, MCP, a server, or whatever works, even by controlling the agent’s own computer. I simply get the result I need without ever opening the application myself.
For example, I’m currently testing — and already using for my business — an accounting ERP system. I haven’t even seen its interface yet.
I’m obviously not using the whole system yet, but at least one business process is already working: collecting invoices and delivery documents from email, Google Drive, and my Amazon Business account → entering them into the ERP and accounting systems → submitting the data to the tax system → generating consolidated reports → and reconciling everything against bank statements, which are also stored on Google Drive.
And I can do all of this through an agent without learning the ERP’s interface.
So the question is: do we really need a UI for every product?
I’m starting to have serious doubts.
Settings are an especially good example. Every new product has its own settings, many of them quite unique. Figuring out manually what each setting does, how it affects the system, configuring everything, and then testing the result can take far longer than simply telling an agent:
“I need it to work like this. Make it happen, test it, and tell me the result.”
I give it the task. I check the result.
Of course, this is still somewhat idealistic. With some products, this approach isn’t possible yet, or doesn’t work completely. But I don’t think this is some distant future. It feels very close.
And looking further ahead, we may not need a separate interface for every product at all.
I already communicate with my agents by voice. For example, I connected Hermes to Telegram, and I’ve been testing it for several days almost entirely through voice. I still haven’t even opened its own interface.
Perhaps the future is a single interface for everything a person needs.
I simply tell the agent what I want. It creates the interface I need for that specific task in five minutes — and then gets out of the way.
At that point, the UI stops being the product.
The UI becomes a temporary tool, created on demand.
0
0
0
23
@a16z 34% paying more just means consumers are trusting an ai that scraped three sponsored reddit threads and hallucinated a coupon code.
0
0
0
21
