To Bing or Not to Bing
Neither! Why not "Bing Me!"? The future of search.
“Oh boy, it’s my lucky day, I get to search!!!”
Said no one, ever.
Outside of some engineers at Google and Microsoft, few people, if any, wake up thinking about search as a fulfilling activity that improves their lives or work.
Instead, people are thinking about something very relevant to them, what they are working on or wanting to do:
“Did any of my sales leads get a promotion, and should I congratulate them?” the salesperson wonders.
“What is the latest AI breakthrough?” asks the software engineer.
“What’s the news for my portfolio companies?” asks the investor.
“What companies are hiring?” asks the college student
“I want to use Siri in my ski helmet; what headphones support this?” asks the work-addicted skier who shall remain nameless in this post…
These are all things that you might use today, such as Google or Microsoft’s Bing search engine, as a tool to help answer those questions. But it often is an unsatisfactory experience—a list of blue links twenty years ago was a fantastic breakthrough to find information on the Internet. But today, it’s not the quickest way to learn something.
If you haven’t had a chance to use the new search functionality in OpenAI’s ChatGPT, sometimes called SearchGPT, it’s incredibly good. Here is a screenshot of a search for ski helmet headphones:
Here is the same search in Google:
And finally, the same search in Bing:
Which of these is better? Note that only ChatGPT actually answered the question directly. The other options provided by Google and Bing don’t support Siri fully. Figuring out the answer to my question would involve a lot of clicking and reading non-relevant web pages.
Almost two years ago, I noticed that my own usage of Google had dropped precipitously with the introduction of ChatGPT:
That trend has only accelerated for me. In reviewing my Google usage for the past month, the number one usage was Google Apps (Google Drive, Gmail, etc.), followed closely by Google Maps. My Google search usage was effectively zero (outside of testing for this article!). I’ve entirely swapped to using AI search from OpenAI and Grok on X.com (formerly Twitter).
I’ve swapped for a simple reason: SearchGPT solves my problem. Ten blue links do not solve my problem—When I have a question, I’m looking for an answer as fast and accurately as possible.
The better mousetrap
Even though SearchGPT is a significant leap forward, very arguably, the fundamental model is still wrong for many scenarios:
Why do I have to search in the first place? Why can’t the computer just tell me what I need to know, when I need to know it?
Obviously, there are times when you have a question and want to know an answer. But for many scenarios, it would be great to have an AI system proactively looking out for you.
Think of many of the scenarios I started with in this blog post. Outside of the shopping one, in each case, an AI could constantly be running in the background and alerting me when something happens.
Broadly speaking, this is the idea behind “Agent-based AI.” Both Salesforce and Microsoft, for example, are talking a lot about agent-based AI approaches. If you aren’t familiar with the concept, this six-minute supercut of Microsoft CEO Satya Nadella’s speech earlier this month at Ignite is well worth watching:
Curiously, one ingredient was missing from the talk: What if my AI agent had the full power of an AI agent combined with the real-time, up-to-date knowledge of a major search engine?
What could be possible?
Microsoft and Google have invested billions in making their respective search engines extremely powerful. Not only have they indexed the web, but they can also find out real-time flight status and the status of your UPS packages and send you alerts when, say, an article is written about you.
Despite those investments and competition between the two technology giants, Microsoft's and Google's relative market share has basically remained unchanged. Search innovation, at least from my usage and experience, has come from elsewhere, like SearchGPT and Perplexity.ai. The “10 blue links” model is not as effective as SearchGPT.
It’s clear that copying Google has not led to a material change in Microsoft’s Bing search engine's competitive market share.
But what if Microsoft took a different approach?
Instead of trying to catch up to Google on the traditional model of doing searches, what if Microsoft instead went all in on the Agentic AI model as the future of Bing search?
It would mean adding real-time search knowledge to the CoPilot experiences Satya talked about above.
Crucially, it would also mean adding real-time web results to their OpenAI API implementation for other developers to use. Currently, this is not available from either Microsoft or OpenAI, even though OpenAI now has SearchGPT (so there is clearly at least an internal API for OpenAI to use).
In a world where magical AI advances seem to happen on a near-weekly basis, it’s tough to get too excited about any particular feature—wait a week, and something else cool will ship! However, a SearchGPT API would be transformative, not just for Microsoft’s Bing but also for nearly every software vendor, particularly business software vendors.
Think about every role within a company, whether it is a sales role, an engineering role, legal, finance, etc. Today, nearly every person in those roles uses search on a reasonably regular basis to perform their jobs—both searching the public web and their internal systems.
Now, imagine an AI agent essentially conducting those searches on its own, learning the information, and synthesizing the results for the employee.
Going back to the customer scenarios I started with, we would have a profoundly different experience for those people:
“Did any of my sales leads get a promotion, and should I congratulate them?” the salesperson wonders. Now, your AI agent can monitor LinkedIn and other news, not only alerting you to relevant changes for your sales prospects but also drafting the start of an email outreach.
“What is the latest AI breakthrough?” asks the software engineer. Now, your AI agent knows the type of software you are working on, and it helpfully and proactively suggests new technologies relevant to your work. If this was then coupled with AI code writing tools like Cursor, Windsurf, v0, Github Copilot, etc., not only could it suggest new technologies to use in that engineers’s work, but it could do or at least start the implementation on its own!
“What’s the news for my portfolio companies?” asks the investor. Many investors have an investment thesis: “We invest in B2B software businesses targeting market verticals.” Now program the AI Agent with that knowledge, and let it alert you when there is relevant news to that thesis—market changes, company announcements, whatever. If the AI agent were deeply integrated into the search index, it could “see” and analyze the same information the search engine was ingesting, analyzing it with the particular insights and investment thesis in mind.
“What companies are hiring?” asks the college student. Same idea here—the student could upload their coursework, their preferences, and so forth, and let the AI continuously scan not just new job postings but also analyze social media posts and so forth to learn who is hiring, what types of questions are asked in the interviews, and so forth. In other words—it could actually help that student land a job versus just showing 100 open positions on LinkedIn that the student would otherwise tediously and mechanically apply to.
I could go on, of course, but I suspect the idea is clear. Search today is a reactive experience driven by humans, starting with a search query. When combined with agentic AI, it could be so much more: a proactive assistant constantly monitoring the world’s knowledge to help each and every person individually with their work and personal needs.
Built as a platform, it could open up a wave of innovation, similar to what App Stores did on mobile phones.
Whether Microsoft, Google, or a new entrant like OpenAI or Perplexity.ai does this, 2025 is going to be another wild year. Moving from a “chat-style” experience to one where AI assistants work on your behalf 24/7 will be profoundly transformative.
Buckle up and Bing Me!