"I don't search anymore, I just ask AI": what that reveals about how we decide what to buy

My client no longer searches Google. He asks AI what to buy. He said it casually, mid-conversation. And as I listened, a question stuck with me: a traditional search engine shows you many results, but AI synthesizes and delivers a few pre-digested recommendations. When it synthesizes, it selects. When it selects, it leaves things out.

From Google to AI: a new way of deciding

For more than two decades, researching a purchase meant roughly the same thing: type something into Google, open several tabs, compare pages, read reviews, visit company websites and form an opinion. The process was tedious, but it had one virtue: the universe of options was visible.

That is changing fast. Today a growing share of consumers prefers to ask an AI assistant directly. According to a McKinsey survey conducted in August 2025, nearly half of consumers already intentionally use AI-powered search engines, and the consultancy projects that by 2028 a large portion of commercial revenue in the United States will flow through this type of search. BrightLocal's 2026 annual survey of local consumers found that the share of people using AI to find recommendations for nearby businesses jumped from just 6% in 2025 to 45% one year later, making it one of the top discovery channels, behind only Google and social media.
Source: McKinsey & Company — New front door to the internet: Winning in the age of AI search

The shift is profound because it is not simply a new tool. It is a different way of deciding. Before, the consumer collected information and chose. Now, more and more, they ask AI to collect, filter and nearly choose for them.

Key idea
A search engine gives you a list so you can choose. AI gives you a nearly finished choice. The convenience is real, but the cost is losing sight of everything that fell outside the synthesis.

The silent bias: what AI doesn't show you

When AI recommends, it does not do so from a complete, neutral market catalogue. It draws on what it learned and what it can find: indexed content, reviews, mentions, structured data, digital presence. A company with abundant content, good positioning and years of online presence is far more likely to appear than a new, small or local company that is still building its digital footprint.

Recent academic research documents this clearly. A 2026 study on generative search, presented at the ECIR conference, found that citations and recommendations from these systems show an exposure bias toward already prominent voices, with the risk of entrenching established players and narrowing the diversity of viewpoints. An earlier audit by researchers at the University of British Columbia had already detected sentiment bias, commercial bias, geographic bias and uneven source quality in the sources these systems favour.
Source: Alipour, Kargar & Zihayat — When Attention Becomes Exposure in Generative Search (ECIR 2026)

One figure summarises the whole problem for me. SOCi's 2026 Local Visibility Index — which analysed more than 350,000 business locations — found that ChatGPT recommended just 1.2% of available locations, compared with 35.9% reached by Google's local pack. In practice, getting an AI to recommend you can be between three and thirty times harder than achieving a good traditional local ranking.
Source: SOCi via Search Engine Land — 2026 Local Visibility Index

The risk is twofold. For the consumer, it is believing they received "the best answer" when they actually received a narrow selection optimised for visibility rather than quality. For small and new businesses, it is becoming invisible — not because they are worse, but because they lack clear, structured, verifiable information where AI looks. This phenomenon already has a name: Generative Engine Optimization (GEO), a field that studies how to get a brand cited by AI assistants, just as SEO sought positioning in Google.
Source: Aggarwal et al. — GEO: Generative Engine Optimization (KDD 2024)

AI as a tool, not an oracle

None of this means using AI for purchasing decisions is a mistake. On the contrary: used well, it is one of the best research tools the average consumer has ever had. AI can organise scattered information, summarise long contracts, compare prices, calculate what something would cost over 12, 24 and 36 months, explain technologies you are unfamiliar with and prepare the questions worth asking a supplier.

The problem is not the tool. It is how you ask.

When you ask AI "which is the best company?", you are pushing it to do exactly what it does worst: deliver a closed verdict based on a selection you cannot see or audit. When you ask "help me compare, without choosing yet", you are using it for what it does best: expanding and organising the analysis rather than closing it down.

How to use AI without falling into biased recommendations

Don't ask: "Which is the best company?"

Ask: "Help me build a matrix to compare several companies. Don't pick a winner yet. I want to evaluate monthly price, installation, contract, technology, maintenance, filters, reviews, exit conditions and total cost over 36 months. Also tell me what data is missing and which small or local companies I should look up on my own."

My case: when I let convenience choose for me

To move beyond theory, I want to ground this in a concrete decision I made myself. I have a Maihue subscription — a purified water service connected directly to the household water supply. The system works, it is convenient and it solves several everyday problems: no carrying water jugs, no buying bottles, no worrying about filter changes and filtered water always available. But when I looked at the service more carefully, the same uncomfortable question I had raised with my client appeared — only applied to me: did I do enough research before signing up?

The honest answer is no.

This case is not intended as an attack on or defence of Maihue. I use it as a starting point to analyse a market that has changed considerably, and to show how a consumer should research before making this kind of decision. Filtered water is no longer simply a product you buy and install. Today it is also a monthly service, with rental equipment, contracts, scheduled maintenance, periodic filter changes and prices tied to a price index. The right question is no longer just "which filter should I buy?" — it is:

Is it better to buy my own purifier, subscribe to a monthly service or keep using water jugs?

Tap water in Chile: regulation versus perception

Before discussing filters, it is worth clarifying something important. In Chile, urban tap water is regulated and monitored. The Superintendencia de Servicios Sanitarios (SISS) publishes drinking water quality results and maintains information on control parameters, maximum limits and sampling conditions.
Source: SISS — siss.gob.cl — Drinking Water Quality Results (verify current link when reading)

This means that when a household subscribes to a filter service, it is not necessarily because the tap water is unsafe. The motivation is often something else: taste, chlorine smell, scale, water hardness, trust, convenience, family habits or simple preference.

This distinction is fundamental because many commercial campaigns in the water market operate in an emotionally charged space — health, purity and safety. The consumer needs to keep two distinct questions separate:

Is the water safe to drink? — This belongs to the regulatory domain.

Do I want or need to filter it? — This belongs to the domain of consumption, convenience and perceived value.

From buying a filter to water as a service

For years the traditional solution was to buy a filter — a jug filter, a tap filter, an under-sink system or a reverse osmosis unit. Then water jugs became popular: purified water arrived in large bottles, placed on a dispenser and replaced when empty. That model solved the taste and availability problem but introduced others: carrying weight, coordinating deliveries, storage space and plastic waste.

The newest model is different: companies that install dispensers connected directly to the mains and charge a monthly fee. Instead of selling only a device, they sell a service: installation, filters, maintenance, support and continuity.

Maihue is one of the most visible brands in Chile in this category. Its proposition is based on mains-connected dispensers without water jugs, with various plans depending on the device type and purification level. The service is billed in UF (Chile's inflation-indexed unit), VAT is included, the contract is annual and installation carries an additional charge per device.
Source: Maihue — maihuechile.cl — Home plans

That raises an important point: when a price is communicated in UF with peso equivalents, separate installation costs and contractual conditions, the consumer needs to convert all of that into real cost. Key questions: How much will I actually pay this month? Does the amount change with UF? Is there an installation fee? Is there a minimum contract? What happens if I want to cancel? Is the equipment mine or on loan?

In my case, two recent invoices show monthly charges close to CLP $20,800. An April 2026 invoice recorded a total of $20,718 and a May 2026 invoice recorded $20,869 for the service "SERVICIO MAIHUE - GREENBIOUF0_437 + IVA". The projected annual spend, if that range holds, approaches $250,000. Over three years it could exceed $700,000, depending on the UF and future adjustments.

The real cost is more than the monthly price

One of the most common mistakes when evaluating subscription services is looking only at the monthly payment. The monthly figure tends to feel small, manageable and easy to justify. But recurring services have a powerful psychological feature: they turn a significant purchase into a permanent drip.

Paying CLP $20,000 per month may sound reasonable. Paying $250,000 per year demands more thought. Paying $750,000 over three years changes the mental frame entirely.

A C+R Research study asked consumers how much they thought they spent per month on subscriptions, then had them add up what they actually paid, service by service. The initial average estimate was around USD $86 per month. The real spend was $219 — more than double. The study found that the vast majority acknowledged it is easy to forget recurring charges.
Source: C+R Research — Subscription Service Statistics and Costs

For water purifiers, a fair comparison should consider at least four scenarios:

Scenario 1: mains-connected monthly service

The user pays a monthly fee. The company installs the equipment, maintains the system and changes filters as agreed.

Scenario 2: buying your own purifier

The user buys the device, pays for installation if needed and then buys replacement filters.

Scenario 3: water jugs or bottled water

The user pays for deliveries or bottles. May require no installation but involves logistics, storage, weight and waste.

Scenario 4: low-cost basic filter

The user buys a jug filter, tap filter or basic under-sink system. Lower upfront cost, but may offer less convenience or require frequent cartridge changes.

Main visible companies and alternatives in Chile

The Chilean market already shows several competing alternatives. Remember the opening warning: this list covers the most visible companies, but good local or smaller alternatives almost certainly exist that do not appear easily in a search — let alone in an AI recommendation.

Maihue

Maihue offers mains-connected purified water dispensers without jugs, with plans for home, business and Horeca. Service is billed in UF with VAT included, annual contract and installation charged per device.
Source: Maihue — maihuechile.cl

Voda Chile

Voda Chile offers purifiers and dispensers under rental or purchase arrangements, with home equipment, installation and technical support. Coverage reportedly includes the Metropolitan, Valparaíso and O'Higgins regions.
Source: Voda Chile — vodachile.cl

AKWA

AKWA is a Chilean company with purified water plans for home and business, including service, maintenance and filter changes according to the chosen plan.
Source: AKWA — akwa.cl

ZeroWater Chile

ZeroWater Chile offers water and coffee solutions for home and office, with options that may include cold, hot, sparkling water or reverse osmosis depending on the equipment or plan.
Source: ZeroWater Chile — zerowater.cl

Culligan Chile

Culligan Chile is primarily oriented towards businesses, offices and commercial spaces. Its mains-connected dispensers include professional installation, filter replacement and sanitisation within rental packages.
Source: Culligan Chile — culligan.cl

Direct retail purchases

Beyond rental companies, there is a broad market for direct purchase: mains-connected dispensers, under-sink filters, reverse osmosis systems, triple filters and jug filters. The key difference between both models is responsibility: when you buy, you take control. When you rent, you pay to delegate.

Convenience as a product

From a marketing perspective, this market is interesting because it does not sell only a technical improvement. It sells friction reduction. Consumers do not always choose the cheapest or technically superior option. They often choose the one that reduces the most mental load.

Buying a filter means researching models, buying spare parts, remembering replacement dates and maintaining the equipment yourself. A monthly service promises to handle all of that. That has value. The question is how much.

A question for the consumer
Are you paying for filtered water, or are you paying to not have to think about filtered water? Both answers are valid. But only one of them is justified by the price, and the other is justified by your time, energy and willingness to take responsibility.

What this market teaches us about software, marketing and AI

For a company like eTips, focused on software development and digital solutions, this case is particularly interesting. The water filter market shows how a traditional industry can transform itself through a logic of service, recurring revenue and experience.

Before, companies sold a product. Now, they manage a relationship. That requires software: client management, contracts, recurring payments, service routes, maintenance scheduling, support tickets, reminders, filter inventory and automated communication.

And there is a new layer that barely existed a couple of years ago: visibility to AI. If more and more consumers ask an AI assistant what to subscribe to, the commercial battle is no longer fought only in search engines or advertisements. It is also fought in whether the AI knows the company, understands it and considers it worth recommending. This changes the rules for everyone — above all for small businesses that may do excellent work yet remain invisible to the system that many consumers now use to decide.
Source: McKinsey & Company — Sign up now: Creating consumer and business value with subscriptions

What to compare before signing up

Before subscribing to a water filter service — or almost any recurring service — it is worth doing a small exercise. It does not need to be perfect, but it should be more thorough than looking at an ad or accepting the first AI recommendation.

First question — technical: What problem do I want to solve? Improving taste is different from reducing scale. Living in a hard-water area is different from one where the main concern is a chlorine smell.

Second question — financial: What is the real cost? Add monthly fee, VAT, price-index exposure, installation, contract, filters, maintenance, spare parts and any cancellation charges.

Third question — behavioural: How disciplined am I? Buying a filter may be cheaper, but only if replacement cartridges are changed on schedule. A poorly maintained filter can become a false sense of security.

Fourth question — contractual: What am I signing? Review minimum commitment, price adjustments, equipment ownership, termination conditions and the actual scope of included maintenance.

Fifth question — trust: Does the company provide verifiable information? A serious supplier should be able to explain what technology they use, what contaminants they reduce, how often filters are changed and what costs are not included.

Sixth question — new: Am I seeing the whole market, or only what AI and search engines showed me first? It is worth actively looking for local or smaller alternatives that may not have appeared in the first response.

Simple comparison matrix

A solid comparison should include these criteria. Even just building this table improves the decision:

A useful prompt for AI-assisted research

A consumer could use a prompt like this to make AI work as an analyst rather than a closed recommender:

I am evaluating whether to subscribe to a water purification service or buy my own purifier. Help me compare alternatives without picking a winner yet. I want a matrix with monthly price, installation cost, whether the price is inflation-indexed or fixed, whether VAT is included, filtration technology, included maintenance, filter change schedule, minimum contract, exit conditions, reputation and total cost at 12, 24 and 36 months. Also tell me what data is missing, what questions I should ask each supplier, and which small, new or local companies I should look up on my own so I don't only consider the most well-known brands.

The last sentence is the most important: it explicitly asks AI to help you counteract its own bias toward the most visible options.

Is Maihue worth it?

The honest answer is: it depends.

It may be worth it if you value convenience, want to avoid water jugs, prefer to outsource maintenance, consume a lot of water and are willing to pay a monthly fee to stop worrying about it.

It may not be worth it if you want to minimise long-term costs, are willing to buy and install your own system and can handle the upkeep yourself.

The important thing is not to decide blindly. In my case, subscribing to Maihue solved a real need, but examining the invoices and the broader market helped me understand what I was actually paying for: not just filtered water, but service, continuity, maintenance and convenience.

Conclusion: AI decides better when we don't let it decide alone

Let us return to the beginning. To my client who no longer searches Google and asks AI what to buy. He was not wrong. AI is an extraordinary tool for researching, organising and comparing. It saves time and processes more data than he could review on his own. His instinct is correct — and it points toward where the world is going.

The problem is not using AI. The problem is confusing a synthesis with the complete truth. An AI that recommends three companies is not showing you the market: it is showing you the part of the market that is most visible to it. And among what was left out there may well be the best option for your situation — that new, small or local company doing great work that is simply not yet visible to the machine.

Artificial intelligence can greatly improve our consumer decisions, as long as we use it as a tool and not as an oracle. Not to choose for us, but to research better, organise more data and ask better questions — while always keeping the final decision ourselves, along with the responsibility to look beyond what we are shown first.

Before subscribing to a water purifier, a software plan, a home alarm, a monthly service or any recurring commitment, every consumer should do one simple exercise: don't just ask how much it costs per month — ask how much it really costs to live with that decision for three years, and make sure you have looked at the whole market, not just what AI decided to show you.

Academic references: Li & Sinnamon — Generative AI Search Engines as Arbiters of Public Knowledge (UBC) · Alipour et al. — When Attention Becomes Exposure in Generative Search (ECIR 2026) · Grossman et al. — How Generative AI Disrupts Search (SIGIR 2026) · Aggarwal et al. — GEO: Generative Engine Optimization (KDD 2024) · BrightLocal Local Consumer Review Survey 2026