Google I/O just changed search as we knew it. But the real lesson is one every IT project manager recognises — what happens when “can we?” replaces “should we?”
Google has quietly (and then all at once) blown up the search engine as we knew it.
At its annual Google I/O developer conference in California, the company unveiled what it calls an “Intelligent Search Box”: a Gemini-powered overhaul that replaces the familiar “ten blue links” with something far more self-contained.
It signals a fundamental shift in what Google Search actually is and reminds Project Managers of a crucial question.
WHAT’S HAPPENED – THE SHORT VERSION
Google no longer wants to send you elsewhere.
Instead of scouring the open web and letting YOU decide what’s worth reading, the new experience is designed to answer, summarise, and engage entirely within Google’s own ecosystem. It’s less a search engine now and more an AI concierge with a search bar attached.
For casual users, that may sound convenient.
BUT for anyone who relies on search to discover independent sources, scrutinise original reporting, or explore beyond algorithmically curated answers, it represents a major shift.
The open web (in all its messy, imperfect, and invaluable glory) just got pushed further from your desktop. Google has spent decades acting as the gateway to the internet, now increasingly seems to want to become the destination itself. Who bought that journey ticket?
This change hasn’t happened in a vacuum, of course. AI has disrupted search behaviour dramatically, with more users turning to tools like ChatGPT instead of traditional search engines entirely. Ironically, one colleague uses ChatGPT to “Google” in the same way you might use a Dyson to “hoover”.
With comparisons being drawn with Nokia, Kodak, and Blockbuster. You can understand why Google felt compelled to respond.
What’s harder to understand is why this was their response.
For most of my life, I’ve known exactly what Google was. Now I’m not entirely sure.
I doubt I’m the only one.
Do you use Google? Did you ask for this?
Then, did your last IT Project’s stakeholders ask for the scope creep?
GOOGLE’S PROBLEM WAS NEVER RELIABILITY
There’s an interesting contrast here with those companies that people now compare Google to: Nokia, Kodak, and Blockbuster.
Those businesses were overtaken by products that were clearly better by the standards of the time. Early smartphones, digital cameras, and streaming services may look primitive today, but compared to phone-bricks, film rolls, and VHS rentals, they were transformational.
AI is different.
AI products have entered the market carrying disclaimers like:
“ChatGPT can make mistakes. Check important info.”
and
“Claude can make mistakes. Please double-check responses.”
That should have been Google’s positioning advantage.
“Our competitors openly admit they’re unreliable and recommend you double-check information… using … Google.”
Instead, Google appears to be leaning further into the very behaviour users remain cautious about.
We never asked Google to curate the information it gave us, not even to fact-check or provide a second source, that was always our job. This change feels like going into a library where the librarian doesn’t let you past her desk, and instead tells you her take on a handful of the books.
THE PROJECT MANAGEMENT LESSON HIDDEN INSIDE THIS
There’s a project management lesson buried in all of this, and it’s one we see repeatedly inside organisations.
A conversation with a client about a project challenge provided a textbook example.
The business commissioned a CRM system with a straightforward objective: help the sales team track leads, manage relationships, and close deals faster. Early versions did exactly that. Adoption was strong. Then the requests started arriving:
Marketing wanted campaign tracking.
Finance wanted forecasting.
Leadership wanted dashboards.
HR suggested onboarding workflows.
Individually, every request sounded reasonable.
Collectively, they transformed the project into something entirely different.
By rollout, the sales team that originally asked for a lightweight CRM found themselves navigating a bloated enterprise platform requiring weeks of onboarding, a dedicated administrator, and endless clicks to perform basic tasks.
On paper, the project succeeded … not massively over budget, just a matter of a few weeks late.
In practice, the people who asked for it lost out.
That’s scope creep in its purest form, and it sits behind a remarkable number of delayed IT projects, budget overruns, and low-adoption systems.
The technology usually works.
The mission just gets lost somewhere between the initial brief and go-live.
WHEN “CAN WE?” REPLACES “SHOULD WE?”
By the time we were called into the project, nobody could clearly articulate what the platform was originally supposed to achieve.
The approved plan and the delivered product were barely recognisable as the same initiative. One colleague described it as “building a jigsaw of a steam train using a box lid with flowers on it.”
Part of the problem was the way AI had been used during development.
The team relied heavily on AI tools to merge every new stakeholder request into the evolving delivery plan. The issue is that AI is extremely good at answering:
“How can we include this?”
It’s far less effective at asking:
“Should we include it at all?”
AI rarely challenges the underlying business need. It doesn’t naturally protect simplicity. It doesn’t defend the original user objective unless a human deliberately keeps that objective in focus.
That’s not an argument against AI.
The potential is enormous.
But I still wouldn’t trust AI with project ROI any more than I’d hire a project manager whose opening line was: “I often make mistakes. Please double-check important information.”
THE WALKMAN LESSON
I’m reminded of the story of the Sony Walkman.
Sony co-founder Masaru Ibuka wanted a portable cassette player so he could listen to opera during long-haul flights.
Sony’s engineers proposed including recording functionality. Technically, it made sense. The additional cost was negligible, and it would add more features.
Ibuka resisted.
Why?
Because it diluted the product.
The Walkman’s appeal was simple: portable music listening. Adding recording capability risked confusing customers about what the product actually was.
It also would have made the device heavier and drained batteries faster — no small issue in the era before USB charging.
Sony went on to sell more than 200 million Walkman cassette players.
The lesson is timeless.
Successful products are not necessarily the ones that do the most.
They are often the ones that remain ruthlessly clear about what they are for.
THE QUESTION THAT MATTERS
Whether you’re delivering an enterprise CRM or redesigning a search engine used by billions, the same principle applies:
Success is not measured solely by delivery dates, feature counts, or budget adherence.
The real question is simpler:
Did the final product actually deliver the thing people originally needed?
That requires judgement, leadership, restraint, and clarity – qualities that still depend heavily on human experience.
AI can help build faster.
It can help merge ideas.
It can help generate options.
But it rarely pauses to ask the question that matters most:
Who asked for this?
And perhaps more importantly:
Who actually benefits from it?
Fortunately, Stoneseed’s PMaaS talent, with their decades of experience, always will.