Google’s New Era of AI Search: The I/O 2026 Updates

Published by David Craig White: 11 July 2026

Estimated reading time: 17 minutes

Google has announced what it describes as a new era for AI Search.

That might sound like standard Google marketing language. But having read through the full announcement, I think this update deserves more attention than the usual product launch.

Google is no longer simply adding AI answers to traditional search results.

It is rebuilding Search around AI.

The new version of Google Search will be able to understand longer and more complex questions, hold conversations, monitor the web for new information, create personalised tools and even take actions on behalf of its users.

In other words, Google Search is becoming less like a list of websites and more like an intelligent assistant.

For businesses, publishers and marketers, that changes the job.

The goal is no longer simply to rank on a results page.

Businesses now need to become trusted sources that Google’s AI systems can understand, retrieve, recommend and use when completing tasks.

The short version

Google announced several major changes to Search at Google I/O on 19 May 2026:

1. Gemini 3.5 Flash is becoming the default model in AI Mode.
2. Google is replacing the traditional Search box with a more intelligent, conversational interface.
3. Users can move directly from AI Overviews into longer conversations.
4. New Search agents will monitor the web continuously for information.
5. Google will expand AI-powered booking and calling capabilities.
6. Search will generate custom tools, dashboards and interactive experiences.
7. Personal Intelligence will tailor Search results using information from connected Google apps.

Google says AI Mode has already passed one billion monthly users. It also says queries in AI Mode have been more than doubling every quarter since launch.

This is not an experiment sitting quietly in Google Labs.

AI search is moving into the centre of Google Search.

Jump to a section

• What Google announced at I/O 2026
• The new intelligent Google Search box
• How Google Search agents will work
• Google Search will take actions for users
• Search results will become custom tools and mini apps
• How Personal Intelligence changes search results
• What these changes mean for SEO
• What businesses should do next
• My view on the future of Google Search
• Frequently asked questions

What did Google announce at I/O 2026?

Google’s announcement centred around a simple idea.

People should be able to ask Google almost anything and have Search work out the best way to help.

That could involve finding information.

But it could also involve monitoring websites, comparing live availability, contacting businesses, producing an interactive tool or using personal information to provide a more relevant answer.

Google is bringing together its traditional search index, Gemini models, real-time data, AI agents and its wider collection of products.

The result is a version of Search that does considerably more than retrieve webpages.

1. Gemini 3.5 Flash is becoming the default model in AI Mode

Google announced that Gemini 3.5 Flash would become the default model within AI Mode around the world.

Gemini 3.5 Flash has been designed to combine speed with stronger reasoning, coding and agentic capabilities.

The word “agentic” matters here.

A normal AI model answers a question.

An agent can work through a task, use tools, check different sources and take actions to help complete it.

By putting a more capable agentic model into AI Mode, Google is preparing Search to handle much more complicated requests.

A user might not search for:

“Best accountant Manchester.”

Instead, they might ask:

“Find me an accountant near Manchester who works with small SaaS businesses, understands international VAT, has good reviews from companies similar to mine and offers an initial consultation.”

That is a completely different type of search.

It contains a location, industry, service requirement, customer preference and commercial intent.

Google’s AI must break the question down, gather information from several sources and decide which businesses best meet the full request.

Google says Gemini 3.5 Flash is now powering AI Mode globally and is designed to support more capable agent and coding experiences.

2. Google is rebuilding the Search box around AI

The traditional Google Search box has remained remarkably familiar for more than two decades.

You enter a few words.

Google gives you some links.

That basic experience is now being redesigned.

Google says the new intelligent Search box will expand dynamically, giving users more space to explain exactly what they need.

It will also offer AI-powered suggestions that go beyond traditional autocomplete.

Users will be able to search using:

• Text
• Images
• Files
• Videos
• Open Chrome tabs

This is important because human questions rarely arrive as perfect keywords.

Sometimes we have a screenshot.

Sometimes we have a document.

Sometimes we have several open tabs and need help comparing them.

Sometimes we simply need enough space to explain the problem properly.

Google is making it easier for people to provide that wider context.

The new intelligent Search box began rolling out in countries and languages where AI Mode was already available following the May 2026 announcement.

What this means for keyword research

Keywords are not disappearing.

People will still search for products, services and information using short phrases.

But keywords will represent only part of the search journey.

More searches will include full questions, detailed requirements, supporting documents and follow-up instructions.

That means businesses must understand the problems behind the keywords.

Knowing that somebody searches for “B2B SEO consultant” is useful.

Knowing why they need one is more valuable.

They may be struggling to appear in ChatGPT.

They may have lost organic traffic.

They may need to generate leads in a new market.

They may be trying to understand why competitors are being cited by AI platforms and they are not.

Content that addresses those deeper needs gives AI systems more useful material to work with.

3. AI Overviews are becoming ongoing conversations

Google is also making it easier to ask follow-up questions directly from an AI Overview.

The conversation can then continue inside AI Mode without the user having to start again.

Google will retain the context from the previous question and use it to make later answers, links and supporting articles more relevant.

This may sound like a small interface change.

It is not.

Traditional search journeys are fragmented.

A person searches.

They open some pages.

They return to Google.

They change the query.

They search again.

Conversational search keeps more of that journey inside a single experience.

The user can refine the question naturally:

“Which of these companies works with small businesses?”

Then:

“Which ones publish clear pricing?”

Then:

“Which one appears to have the strongest experience in my industry?”

Google remembers what “these companies” means.

The user does not need to repeat the original query.

Search intent will become clearer as conversations continue

This gives Google much more information about what the user actually wants.

The first search may be broad.

The second adds a preference.

The third introduces a commercial requirement.

The fourth may lead to an enquiry or booking.

The search journey becomes progressively more specific.

This creates a strong opportunity for businesses with clear positioning.

If your website makes it obvious who you help, what you offer, where you operate and why somebody should choose you, Google has a better chance of matching your business to a detailed request.

Vague businesses will become harder to recommend.

Clear businesses will become easier.

4. Google is introducing Search agents

Search agents are probably the most important part of the announcement.

Google says users will be able to create and manage multiple AI agents directly within Search.

The first will be information agents.

These agents will operate continuously and monitor information across sources including websites, news publishers, blogs and social media.

They will also be able to use fresh Google data related to areas such as finance, shopping and sport.

A user could ask an agent to:

• Monitor apartment listings matching detailed requirements.
• Watch for changes in prices.
• Track new product releases.
• Follow updates from selected companies.
• Find new research about a particular subject.
• Alert them when specific conditions are met.

Instead of conducting the same search every few days, the user gives Google a standing instruction.

The agent then searches on their behalf.

Google said information agents would launch first for Google AI Pro and Ultra subscribers during summer 2026.

Search is moving from individual queries to persistent instructions

This changes how I think about search demand.

Until now, we have generally measured search as a sequence of individual queries.

One person performs one search.

An agent may repeatedly check the web hundreds of times in response to a single instruction.

That creates a new type of search activity.

Businesses will not only be trying to appear when a person searches.

They will also be trying to become a source used by an agent that continuously gathers information.

That places greater importance on:

• Accurate information
• Clear publication dates
• Regular updates
• Consistent business details
• Accessible page structures
• Strong product and service descriptions
• Original evidence
• Reliable third-party mentions

An agent needs to know whether information is current.

A page that was useful three years ago may not be useful today.

Freshness will not matter equally for every subject, but it will become increasingly important for news, research, pricing, product availability, services and fast-moving industries.

5. Google Search will take actions for users

Google Search agents will not stop at gathering information.

Google is also expanding the ability of Search to help users complete tasks.

This includes finding real-time availability for bookings and sending users directly to suitable providers.

Google has already introduced agentic booking features across areas such as restaurants, event tickets, beauty appointments and wellness services in the United States.

Google also announced plans to expand these capabilities into more local services.

For selected categories, including home repairs, beauty and pet care, users in the US will be able to ask Google to call businesses on their behalf.

Google has already tested a version of this calling feature where Search contacts nearby retailers, asks questions and sends the user a summary of what it finds.

This could change local search dramatically

Imagine somebody needs an emergency plumber.

Instead of opening several websites and calling each company, they could ask Google:

“Find a plumber near me who can repair a leaking boiler today, has strong customer reviews and charges less than £150 for an emergency call-out.”

Google could identify suitable companies.

It could check their business information.

It could potentially contact them.

It could then present the best available options.

The customer has not visited ten websites.

They may not have visited any.

The winning business could be the company that gave Google the clearest, most reliable and most useful information.

That means local optimisation will increasingly involve more than appearing in a map pack.

Businesses will need complete Google Business Profiles, accurate service information, good reviews, current opening hours and clear ways for systems to check availability or complete a booking.

Being visible will matter.

Being usable will matter more.

6. Search results will become custom tools and mini apps

Google is bringing agentic coding capabilities into Search through Gemini 3.5 Flash and its Antigravity technology.

This means Search will be able to generate a custom interface based on the user’s question.

Instead of showing the same style of results page to everybody, Google could create:

• Interactive tables
• Calculators
• Graphs
• Simulations
• Visual explanations
• Comparison tools
• Dashboards
• Trackers

Google says these generative interface capabilities will be made available to users for free.

More advanced persistent tools and mini apps are expected to reach paid subscribers first.

Google gives the example of creating a personal fitness tracker that uses live information such as local weather, maps, reviews and nearby services.

But the wider implications are enormous.

A user might ask Search to build:

• A mortgage comparison dashboard
• A house-moving checklist
• A software vendor comparison
• A wedding budget tracker
• A personalised learning plan
• A project management tool
• A holiday planning dashboard

Google could pull information from several websites and organise it into something built specifically for that user.

The search results page itself is becoming dynamic

This means businesses may no longer appear in a neat list of ten blue links.

Their information could appear inside a comparison table.

Their prices could feed into a calculator.

Their service details could become part of a personalised dashboard.

Their research could support a graph or generated explanation.

Their product information could be used inside a recommendation tool.

The content still matters.

But Google may change the format completely.

Businesses therefore need to publish information in a way that can be extracted, understood and reused accurately.

7. Personal Intelligence will make Search results more individual

Google is also expanding Personal Intelligence within AI Mode.

With permission, users can connect services such as Gmail and Google Photos, with Google Calendar expected to follow.

Search can then use that private context to provide more relevant answers.

For example, it could recommend products based on previous purchases or build travel plans around existing bookings and past trips.

Google says users decide which services they connect and can turn those connections off.

Two people may receive completely different answers

Search results have been personalised for years through factors such as location, language and search history.

Personal Intelligence takes this much further.

Two people could ask exactly the same question and receive different recommendations because Google understands more about their individual circumstances.

This makes the idea of one universal search result increasingly outdated.

Businesses may rank well for one type of customer but not another.

That is not necessarily a problem.

The real goal is to become highly relevant to the people most likely to buy from you.

Strong positioning will help AI systems understand that relevance.

What do Google’s AI Search updates mean for SEO?

I do not believe these announcements mean SEO is dead.

People have been declaring SEO dead for about as long as SEO has existed.

But I do believe the definition of SEO is expanding.

Traditional SEO asks:

“How do I get this page to rank for this keyword?”

Modern AI search optimisation asks:

“How do I make this business the most credible and relevant source for this subject, customer and situation?”

Those are related questions.

But they are not identical.

1. Rankings will no longer tell the whole story

A business could rank well in traditional results while being almost invisible inside AI answers.

Another business may receive fewer traditional clicks but be repeatedly mentioned, cited or recommended by AI platforms.

Businesses will need to monitor a wider collection of signals, including:

• Traditional keyword rankings
• Organic website traffic
• AI Overview visibility
• Brand mentions in AI answers
• Source citations
• Referral traffic from AI platforms
• Google Business Profile actions
• Leads generated without a recorded website visit

Search visibility will become more fragmented.

Reporting must catch up.

2. Clear positioning will become an SEO advantage

AI systems need to understand what a company does and when it should be recommended.

A website that says:

“We deliver world-class solutions that empower modern organisations.”

provides almost no useful information.

A website that says:

“We provide B2B SEO and AI search consulting for SaaS companies selling into the UK and European markets.”

is much easier to understand.

Specific language helps users.

It also helps machines.

Businesses should clearly explain:

• What they sell
• Who they help
• Where they operate
• What makes them different
• Which problems they solve
• Which industries they understand
• How customers can buy or make contact

3. Content must answer complete questions

Short pages written around one keyword may struggle to satisfy complex AI search requests.

That does not mean every page needs to contain 5,000 words.

It means the content should answer the important questions properly.

A good service page should not only name the service.

It should explain who it is for, what is included, how the process works, what problems it solves, how much it costs where appropriate and what results a customer can reasonably expect.

Useful depth beats empty length.

4. Original information will become more valuable

AI systems can summarise generic information easily.

There are already thousands of articles explaining basic topics such as SEO, email marketing or sales prospecting.

Repeating the same advice with slightly different wording adds little value.

Original information is harder to replace.

That could include:

• First-hand experience
• Customer research
• Surveys
• Experiments
• Case studies
• Expert analysis
• Unique data
• Product testing
• Strong opinions supported by evidence

The more original value a business publishes, the more reasons an AI system has to use or cite it.

5. Third-party authority will matter

A company cannot build authority entirely through claims made on its own website.

Other sources help confirm what a business is known for.

These may include:

• Industry publications
• Customer review platforms
• News websites
• Trade associations
• Business directories
• Partner websites
• Podcasts
• Research reports
• Expert round-ups
• Relevant online communities

This is one reason digital PR, reputation building and brand mentions are becoming closely connected to AI search optimisation.

AI systems do not only need to find your claims.

They need reasons to trust them.

6. Live business information will become more important

Agentic Search needs accurate information to make recommendations and complete actions.

For local businesses, this includes:

• Opening hours
• Service areas
• Appointment availability
• Pricing
• Stock information
• Contact details
• Customer ratings
• Booking options

Incorrect or outdated information creates risk for Google.

If Google recommends a closed business, an unavailable service or an incorrect price, the user loses trust.

Businesses that maintain accurate information will be easier for agents to use.

What should businesses do now?

I would not recommend rebuilding an entire website every time Google announces a new AI feature.

That is a good way to waste money.

The fundamentals still matter.

But I would start adapting the way content and business information are managed.

1. Make your business easy to understand

Use plain language.

Say exactly what you do.

Avoid vague marketing claims.

Create clear pages for your main services, products, industries and locations.

2. Answer the real questions customers ask

Speak to customers.

Review sales calls.

Read support messages.

Look through relevant forums and communities.

Find the questions people ask before they buy.

Then answer those questions clearly on your website.

3. Keep important information current

Review pages containing prices, availability, product details, statistics and recommendations.

Add visible publication and update dates where they are useful.

Remove information that is no longer correct.

4. Strengthen your wider online presence

Make sure your company information is consistent across your website, Google Business Profile, social accounts, directories and review platforms.

Build genuine relationships with relevant publishers and industry organisations.

Give people a reason to talk about your business.

5. Publish information worth citing

Do not simply rewrite what everybody else has already published.

Add experience.

Add evidence.

Add examples.

Run small studies.

Share what you have learned.

Useful original information gives both people and AI systems a reason to remember you.

6. Prepare for actions, not just visits

Make it easy for customers and platforms to complete the next step.

That might include:

• Requesting a consultation
• Checking availability
• Booking an appointment
• Viewing prices
• Comparing products
• Calling the business
• Getting directions
• Completing a purchase

The future of search is not simply finding an answer.

It is getting something done.

My view: Google Search is becoming an operating system for decisions

The most important part of this announcement is not Gemini 3.5 Flash.

Models will continue to change.

There will be another Gemini version, another interface and another collection of features.

The real shift is that Google is turning Search into an operating system for decisions.

It can understand what somebody needs.

It can research the options.

It can monitor information over time.

It can organise the results.

It can create a tool.

It can contact providers.

It can help the user act.

That shortens the distance between a question and a decision.

For businesses, there is an uncomfortable truth here.

Some website visits will disappear.

People will sometimes get what they need directly from Google without clicking through to several websites.

But businesses will not become irrelevant.

Google still needs sources, products, services, experts, evidence and providers.

The opportunity is to become one of the businesses Google trusts enough to use.

That will require more than inserting keywords into pages.

It will require clear positioning, useful content, strong reputation signals and accurate information across the web.

SEO is not disappearing.

It is becoming part of something larger.

That something is AI search.

Frequently Asked Questions

Source

This article is based primarily on Google’s official announcement, “A new era for AI Search,” published on 19 May 2026, alongside related Google I/O 2026 product announcements.

About the Author

Hello – I’m David Craig White, an independent AI search consultant with over 20 years’ experience in SEO, B2B SaaS, positioning and commercial growth.

When I’m not writing articles about AI search, I’m helping B2B companies improve how they are found, understood and recommended across Google, ChatGPT, Claude, Gemini, Perplexity and other AI answer engines.

Read more about my products and services below:

👉 Subscribe to get my latest updates here

I also post regularly on LinkedIn — Feel free to connect.

Page last updated: July 2026