Search Surface Changes

Search product managers at Google, Microsoft, and everywhere else refer to the different search interfaces as “surfaces,” and hoo boy, have the surfaces been shifting?

➣ Microsoft is rolling out CoPilot on Mac a desktop version of the Copilot chatbot that I’ve honestly never quite understood as it seems like ChatGPT but with extra steps.

➣ Amazon’s making Alexa on Web so you can get to it from anywhere, which is maybe useful? (more on this in a second)

➣ Meta is rolling out a dedicated MetaAI Mobile App

Deep Research Overview

It’s becoming increasingly clear that the first mainstream AI “Agent” is going to be the Deep Research tools that all the AI and search companies are making.

I view this generally as a positive as we all already have a good handle on what makes a quality response from traditional search so:

• They’re easy to reason about and evaluate
• They’re immediately helpful to a much larger group of people

Here’s a practical example. I was looking for an episode of the Software Social podcast where they spoke about virtual assistants.

Consider:

• A traditional Google search for software social podcast virtual assistant episode turned up tons of noise.
• ChatGPT (4o) gave me a generic non-answer about using VAs in podcast production.
• ChatGPT Deep Research took 5 minutes, went through 24 different sites, and in the end, perfectly nailed the answer.

AI Mode is Deep Research Done Quick

AI Mode is Google’s new term for mixed AI + traditional search products that are slowly being released to everybody.

It’s interesting for a couple of reasons:

1. In many ways it’s a faster version of the “Deep Research” mode that all the different AI services have been pushing. In the example above, it took 5 minutes to get an answer (time I could have spent productively watching YouTube videos of how to restore French millwork or something).
2. It’s so boring and immediately useful. In the same way that most people using Google for search don’t really realize that they’re getting an AI overview from some system called Gemini, people are just going to use it and shrug “I guess I got my answer”.

Software Spotlight Podcast

I got to speak about “Revolutionizing SEO and AI Search Strategies” with Michael Bernzweig this week on the Software Spotlight podcast.

One of the things we cover is the changing nature of “intent” in SEO and how we now need to consider both “task intent” as well as “search intent”.

Software Spotlight Episode 121

Michael also publishes the Tech Spotlight newsletter, which covers many issues in the B2B consulting world that are extremely relevant to SEO freelancers and agencies. I’d recommend you check it out.

Knowatoa + Gemini 2.0 Flash

We’ve been impressed with Google’s new Gemini 2.0 Flash model, and we’re excited to announce that right now if you’re a paying Knowatoa customer, you can log in and see how your brand ranks in Google’s latest.

Google has publicly stated that AI overviews, AI mode, and DeepResearch are all using versions of Gemini, making it crucial for your search in the future to start tracking and optimizing for it.

Thanks

I had some excellent conversations on LinkedIn this week with Suganthan Mohanadasan on about how much “AI Optimization” is just “keep doing the same SEO stuff” and with Kevin Indig on Face vs No-Face Content Strategy both good reads and good people.

See you next week,

Mike

p.s. It would really help me out if you could Follow me on LinkedIn