How AI Is Replacing The Search Bar

Ada  00:00

If you are a brand marketing product today, there’s a very high chance you are spending 1000s, or I mean maybe millions of dollars optimizing for a traditional search bar that a massive chunk of your future customers has completely abandoned.

Leo  00:15

Yeah, they’re just entirely gone.

Ada  00:16

Right. Welcome back to the deep dive. We are thrilled to have you, you know, you the endlessly curious learner with us today, because we are looking at an incredibly revealing stack of internal strategy documents and positioning decks. These are from Accenture and DaVinci Commerce,

Leo  00:34

and what these documents reveal is this fundamental, almost invisible rewiring of how discovery actually happens on the internet right now.

Ada  00:43

Yeah, it’s wild. It

Leo  00:44

really is a quiet revolution. I mean, we’re watching the manual keyword-driven search process just collapse in real time, and it’s being replaced by what the industry is calling agentic commerce. Basically, conversational AI is taking over the very top of the shopping funnel.

Ada  00:57

Okay, let’s unpack this because to really understand what DaVinci Commerce is doing, and they have the specific magic trick we’re going to get into in a bit. We first need to grasp the sheer staggering scale of the problem brands are suddenly waking up

Leo  01:11

Oh, absolutely!

Ada  01:12

I was reading through the Accenture data in our sources, and the numbers actually made me pause. Chat GPT alone currently has 900 million weekly active users,

Leo  01:23

which is huge,

Ada  01:24

right? And they are handling 2.5 billion prompts a day. 2.5 billion.

Leo  01:31

Yeah, and just to put that 2.5 billion into perspective, that volume is already roughly 30% of Google’s entire search volume.

Speaker 3  01:38

Wow. We

Leo  01:39

are watching a tectonic shift in traffic, but you know the more critical data point in these sources: who is driving this shift and how it alters their behavior.

Ada  01:46

Right, the demographic

Leo  01:47

exactly. 80-5% of Gen Z is now using AI for shopping discovery. They prefer it over traditional search engines. They prefer it over like social media scrolling, and they even prefer it over human influencers.

Ada  01:58

Which I completely understand. Honestly, I mean, think about the traditional search engine experience today. I was looking for a new espresso machine last week. I typed it in, and the results page was just a massive, disorganized warehouse. You get 14 sponsored links, a bunch of SEO stuffed articles that don’t actually answer your question, and endless pop-ups.

Leo  02:20

It’s exhausting.

Ada  02:21

It is. You suffer from intense decision fatigue before you’ve even read a single spec. Conversational AI shopping, by contrast, is like walking into a high-end boutique and sitting down with a highly competent personal shopper.

Leo  02:34

That’s a great way to put it.

Ada  02:35

You just explain your exact lifestyle, and they hand you three perfect options.

Leo  02:39

That is the exact psychological shift, and the data proves it works better. Yeah, when shoppers use AI, the results are undeniable. I mean, AI influenced retail traffic is seeing a 40-2% year-over-year conversion left.

Ada  02:52

That’s massive.

Leo  02:53

It is, and even wilder, their basket sizes-you know, the total amount they actually purchase at checkout. Those are 30-5% bigger compared to non-AI shopping.

Ada  03:01

30-5% bigger. That means people aren’t just finding what they want faster. The AI is actually earning their trust to the point where they feel confident buying complementary items.

Leo  03:10

Exactly. But

Ada  03:10

here is where the plot thickens for the brands themselves. As a brand, you look at those conversion numbers and say, “Well, great, get me in front of that AI shopper.

Leo  03:19

Sure, obviously.

Ada  03:20

But according to these strategy decks, the way brands are currently trying to do this is failing spectacularly,

Leo  03:28

utterly failing, because well, brands are trying to send their warehouse inventory list to that high-end personal shopper.

Ada  03:34

Oh, I see.

Leo  03:35

Currently, brands rely on what they call PM data, product information management. This is the centralized database where a brand keeps all its technical specs, weights, dimensions, materials, all that.

Ada  03:47

Right, the dry stuff.

Leo  03:48

Yeah. So when they plug that traditional data feed into an AI, they are basically handing the AI a dry technical spec sheet.

Ada  03:56

So they are feeding the AI the exact same bullet points they’ve been pasting onto the bottom of a web page for the last 15 years.

Leo  04:03

Yes, and to understand why that doesn’t work, we have to look at the mechanics of how a large language model or an LLM actually functions. Right, LLMs, which are the underlying brains powering chatbots like ChatGPT or Claude, they don’t think like a traditional database. They predict language based on vast amounts of human context.

Speaker 4  04:23

Okay.

Leo  04:23

What’s fascinating here is how the sources provide a really clear example of this disconnect using a smartphone, the Nexor Blaze x1

Ada  04:32

Okay, let’s hear it.

Leo  04:33

So the brand’s product feed lists features like a 6.78 inch 144 hertz AMOLED display, vapor chamber cooling 2.0 and a 6,000 mil or battery.

Ada  04:43

Okay, so imagine a shopper goes to ChatGPT and types, “I need a good phone because I play daily mobile games for about four to five hours on my commute.

Leo  04:50

Right.

Ada  04:51

My guess is the AI isn’t going to pull up the Nexora, even though it has a massive battery and a cooling chamber.

Leo  04:56

It absolutely won’t.

Speaker 5  04:58

Yeah.

Leo  04:58

Because the AI is looking for human. Context. If the LLM hasn’t been explicitly trained to connect the phrase vapor chamber cooling 2.0 with the human experience of playing games for five hours without the phone getting hot in my hands,

Ada  05:09

it just misses it completely.

Leo  05:10

Exactly. It cannot bridge the gap. The AI wants raw data, yes, but it desperately needs intent-based context. It needs to understand the human application of those factory specs.

Speaker 6  05:22

Right.

Leo  05:23

Without that translation, the brand is effectively invisible inside the AI ecosystem.

Ada  05:29

It’s like a language barrier. The AI is saying, “I speak human need, and the brand is screaming back in factory specs.

Leo  05:36

That’s exactly it.

Ada  05:37

So, if basic spec sheets fail that human context test, how does a brand actually fix it? Because nobody has the time to manually rewrite 10,000 product descriptions.

Leo  05:48

No, of course not.

Ada  05:48

Which brings us to DaVinci Commerce and what they call their content enrichment engine.

Leo  05:54

And this engine is a masterclass in solving a complex data problem without reinventing the wheel. Da Vinci realized that the human context already exists? It’s just scattered all over the internet.

Ada  06:03

Right. Reading through the breakdown, the content enrichment engine takes that dry PIN data-you know, the battery size, the fabric type-and it systematically layers it with real-world context.

Leo  06:13

Layering is the key word there.

Ada  06:15

Yeah, they pull in verified bizarre voice reviews, which gives the AI access to star ratings, customer images, and the actual unvarnished text of what verified buyers are saying-huge

Leo  06:28

data source.

Ada  06:28

Totally. Then they pull in UGC, user-generated content, meaning they are scraping the raw, honest discussions from Reddit threads and YouTube video transcripts.

Leo  06:39

Yeah.

Ada  06:39

And finally, they layer in search trends from tools like Semrush, so the engine knows what specific colloquial keywords people are actively searching for this month.

Leo  06:48

They are acting as a real-time translator at the United Nations. They take the technical language of the product, combine it with the conversational language of the consumer, and synthesize it into a format the LLM can instantly understand and recommend.

Ada  07:01

It’s like taking a flat black and white 2d photograph of a product and turning it into a 3d holographic movie.

Leo  07:07

Yeah, that’s a perfect analogy.

Ada  07:08

It changes the data from what is this made of to how do actual human beings use this in the real world.

 

Leo  07:14

Exactly. I

 

Ada  07:15

do have to ask though. Pulling in Reddit threads and YouTube comments sounds incredibly risky.

Leo  07:19

Oh, absolutely.

Ada  07:20

I mean, if I’m a brand manager, I know that Reddit can be a complete cesspool of sarcasm and trolls. If Da Vinci is feeding raw Reddit UGC into an AI to train it on my product, how do they ensure the AI doesn’t start recommending my product based on a sarcastic joke?

Leo  07:36

That is a phenomenal question, and it gets to the heart of how the engine actually works. It isn’t just doing a blind copy paste of the internet.

Ada  07:44

Okay, good. The

Leo  07:44

content enrichment engine uses its own layer of AI to parse that user-generated content for sentiment and relevance. It filters out the noise, the sarcasm, the irrelevant tangents.

Ada  07:56

So it’s smart enough to know when someone is kidding.

Leo  07:58

Yes, it identifies patterns. Like if 50 different people on Reddit genuinely mention that a specific hiking boot has a narrow toe box, the engine captures that verified contextual truth and layers it onto the product data.

Ada  08:10

That makes so much more sense. It’s distilling the truth of the crowd.

Leo  08:13

Precisely.

Ada  08:14

Let’s look at the specific example the Da Vinci strategy decks used to illustrate this because it clicked instantly for me. They use a skincare product, specifically a dark circle moisturizer.

Leo  08:24

Oh, that’s a great example.

Ada  08:25

Yeah, a shopper opens up their AI agent and types suggest a moisturizer to treat my dark circles. Now, the brand’s official PM feed might only list the chemical compounds: hyaluronic acid, peptides, ceramides.

Leo  08:38

Right, the science.

Ada  08:39

Right, it doesn’t actually contain the colloquial phrase “dark circles” anywhere in its official marketing text.

Leo  08:45

So, under the old model, the AI evaluates the prompt dark circles, scans the brand spec sheet, finds zero matches, and just moves on to a competitor.

Ada  08:54

Exactly, that’s precisely the blind spot. But with Da Vinci, their content enrichment engine has already gone out and found a verified Walmart review where a customer literally wrote, “This moisturizer works super well for my dark circles.

Leo  09:06

There’s the human context.

Ada  09:07

Yes, the engine binds that human review to the technical spec sheet and feeds the whole enriched package to the AI.

Leo  09:15

And suddenly, the LLM has everything it needs. It understands the user’s intent. It sees the verified human experience matching that intent, and it cross references the chemical ingredients to ensure it’s a valid recommendation. The brand becomes perfectly discoverable.

Ada  09:29

It’s incredibly resourceful. I mean, they are taking the unpaid labor of the brand’s own customers, the people writing the reviews and making the YouTube tutorials, and using their voices to translate the product for the AI,

Leo  09:42

and this enriched data forms the foundation of what Da Vinci calls an agentic brand store. It is no longer just a static product listing. Right. It becomes a fully branded, highly conversational shopping experience that lives natively inside the AI interface.

Ada  09:58

Okay, but this sparks a huge. Operational dilemma. Once you have this beautifully enriched data, where do you put it? From everything we are reading here, relying on ChatGPT as your only distribution channel is a terrible idea.

Leo  10:10

Well, it would be a fatal error strategically. Yeah, the sources highlight a critical piece of modern consumer behavior, which is extreme shopper fragmentation.

Ada  10:18

People are everywhere.

Leo  10:19

Exactly, the data shows that 79% of consumers who actually pay for premium AI tiers pay for both ChatGPT and Claude.

Ada  10:27

Both of them.

Leo  10:28

Yes, they are completely platform agnostic. They treat these AIs like utilities. They will use whatever agent is open on their screen or integrate it into their phone at that exact moment.

Ada  10:39

People aren’t loyal to a specific search engine anymore. They just want the answer

Leo  10:42

precisely. Because of this fragmentation, Da Vinci designed the Agentic brand store to be what they call Omni LLM.

Ada  10:50

Okay.

Leo  10:50

It pushes this enriched context-heavy data out to everywhere simultaneously. It goes to ChatGPT, to Google’s Gemini, to Anthropics Claude.

Speaker 4  11:01

Wow!

Leo  11:01

It pushes to retailer-specific agents like Alexa for shopping, Walmart’s Sparky, and Target’s AI assistant. It even feeds the brand’s own internal website chatbot. The strategy is to be omnipresent wherever the consumer decides to start a conversation.

Ada  11:17

Let me play devil’s advocate here, though. Sure. If I’m the chief technology officer at a massive enterprise brand listening to this. I’m thinking, well, we have hundreds of developers on payroll. We have our own data. Why can’t we just build this ourselves?

Leo  11:28

That’s the classic question,

Ada  11:29

right? Just hire some engineers. We plug into the OpenAI API, you know, the digital bridge that lets our servers talk directly to theirs, and we’re done.

Leo  11:38

It’s the logical first thought for any engineering team, but the DaVinci documents provide a very harsh reality check on the build-it-yourself trap.

Speaker 7  11:47

I bet

Leo  11:48

the first issue is time to market. Building a custom AI commerce agent from scratch, engineering the prompt structures, and integrating the enriched content takes a massive enterprise brand anywhere from six to 12 months.

Speaker 8  12:01

Wow!

Leo  12:02

And in the current AI landscape, 12 months is a geological era. The technology will have iterated three times before you launch.

Ada  12:08

That is entirely fair. But time aside, what’s the technical hurdle?

Leo  12:12

The technical hurdle is the user interface or UI. If you build it yourself, your engineers have to manually resize and transform your brand’s assets. You know the images, the text formatting, the tables for every single LLM interface. Oh,

Ada  12:25

I see.

Leo  12:25

Yeah, ChatGPT displays a product comparison table very differently than Claude does. If you don’t build custom formatting for each one, your UI breaks and your brand looks terrible to the consumer.

Ada  12:35

Oh wow! I didn’t even think about the visual formatting.

Leo  12:38

And the third, and perhaps most dangerous risk of building it yourself is vendor lock-in. Let’s say you spend a year building a brilliant custom integration specifically for OpenAI’s ecosystem. What happens if Google’s Gemini suddenly dominates the market next year, or what if OpenAI changes their core API protocols tomorrow?

Ada  13:00

You’ve basically built a highly customized, expensive railroad track for a specific type of train that just went out of business. That

Leo  13:07

is a much better way to look at

Speaker 9  13:08

Yes,

Leo  13:10

Da Vinci, conversely, is turnkey. They act as the universal adapter.

Ada  13:14

That makes sense.

Leo  13:14

The sources point to a pilot program they ran with Diageo for the Tanqueray brand. It’s the perfect case study for speed.

Ada  13:21

Okay, tell me about Tanqueray.

Leo  13:22

So Da Vinci ingested over 70 different products and complex cocktail recipes, ran them all through the content enrichment engine, navigated all the strict alcohol compliance and age gating rules, and had a fully functioning agentic brand store live on ChatGPT in less than two weeks.

Ada  13:38

Two weeks versus 12 months-that is wild. Yeah, but wait, moving 70 products of corporate data into ChatGPT in two weeks triggers massive alarm bells for me. How so? My immediate thought as a lawyer or a compliance officer is that this is a data privacy nightmare. Are these enterprise brands really just taking their proprietary internal data? You know, their unreleased product specs, their high-res imagery, their internal CRM data showing customer purchase history, and throwing it out into the open web for OpenAI to scrape and train on.

Leo  14:11

Not at all, and this is a crucial distinction made in the Accenture materials. DaVinci acts as an enterprise protection layer.

Ada  14:16

Okay,

Leo  14:17

they utilize a mechanism often referred to in the industry as retrieval augmented generation or UA, but the simple way to understand it is like a high security vault with a glass window.

Ada  14:27

Okay, walk me through the vault.

Leo  14:28

Each brand’s proprietary data is isolated and tenanted securely within the DaVinci platform, the vault. Right. When a shopper asks a question on ChatGPT, the AI comes to the window. It asks DaVinci for the answer. Da Vinci’s answer agent securely retrieves only the relevant, legally approved, enriched information and slides it through the window in real time to answer the shopper. The AI models are explicitly restricted by contract from ever entering the vault and using any of that proprietary brand data for their own foundational model training.

Ada  14:58

That is massive for. Teams. It’s the difference between having a conversation through the AI rather than giving your brain to the AI.

Leo  15:05

Precisely, and it highlights a very sharp contrast between active curation and what the industry calls passive GEO or generative engine optimization.

Ada  15:15

GEO is everywhere right now. Every marketing agency is selling GEO services.

Leo  15:19

They are, but GEO is inherently passive. It relies entirely on making your public-facing website easily readable, so that when an AI crawler wanders by and scrapes the public domain, it hopefully understands your site better.

Ada  15:31

Hoping is the key word there.

Leo  15:33

Right. It’s static. It’s heavily reliant on text, and most importantly, you have absolutely zero control over how the AI ultimately presents your brand alongside your competitors,

Ada  15:42

which is terrifying for brand consistency.

Leo  15:45

Exactly, Da Vinci’s Agentic Brand Store is active. It is brand proprietary, highly curated, and provides a direct immersive engagement where the brand controls the visual and conversational experience.

Ada  15:58

Okay, so we’ve unpacked the tech, the speed of deployment and the legal safety mechanisms. Right, but let’s get down to brass tacks here. What is the actual financial impact of doing this right? Because at the end of the day, a CTO needs to justify the investment.

Leo  16:11

The ROI models outlined in these documents are very explicit. They detail the three levers of agenda commerce value. In total, DaVinci drives a verified 6.5% total digital revenue uplift for a brand.

Ada  16:24

A 6.5% uplift in total digital revenue just by changing how an AI perceives your data.

Ada0  16:30

Yes,

Ada  16:30

I mean for a billion-dollar enterprise brand, that is 10s of millions of dollars. How exactly does that break down?

Leo  16:35

It breaks down into three distinct financial levers. Lever one is rescuing revenue at risk. Right now, brands that are not properly enriched for LLMs are literally losing market share to competitors who are. Da Vinci estimates that plugging into their engine rescues a 3% loss in digital revenue that was bleeding out from consumers abandoning traditional Google search.

Ada  16:56

So the first 3% is literally just stopping the bleeding.

Leo  16:59

Exactly.

Ada  17:00

What’s lever two?

Leo  17:00

Lever two is generating net new revenue by creating these highly personalized, conversational, lifestyle-driven discovery experiences. Brands attract new shoppers who convert at a much faster rate.

Ada1  17:13

Makes sense.

Leo  17:14

That accounts for a 1.5% increase,

Ada  17:15

which leaves 2% for the final lever.

Leo  17:18

Correct. And lever the three is driving higher basket values because the AI acts like that brilliant personal shopper we talked about earlier. It makes highly contextual, use case-based multi-product recommendations. Oh right! If you buy that dark circle moisturizer, the AI intuitively cross-references user reviews and suggests the exact right complementary FPF that won’t pill under makeup. The consumer buys both. That drives another 2% in digital revenue.

Ada  17:44

Stop the bleeding for 3% Capture new conversions for 1.5% and upsell intuitively for 2%

Leo  17:51

There’s for six and a half percent.

Ada  17:52

So what does this all mean? It really hammers home the absolute necessity of not being left behind in this AI commerce gold rush. If you aren’t actively managing how these language models perceive, translate, and recommend your products, you are leaving millions of dollars on the table for a more agile competitor to scoop up.

Leo  18:10

It is no longer a futuristic abstract concept.

Ada2  18:13

Yeah,

Leo  18:14

the shoppers are already there. They’re already having these conversations, and they are already spending money.

Ada3  18:18

Yeah,

Leo  18:18

the only question is whether your brand possesses the human vocabulary required to participate in the chat,

Ada  18:24

and what a journey we’ve been on today to understand the mechanics of that vocabulary. We started by looking at the staggering numbers behind the death of the traditional search bar. We unpacked exactly why basic dry PM data and spec sheets make your brand completely invisible to an LLM, and we took a deep look at the solution: how Da Vinci’s content enrichment engine acts as a translator, taking those basic feeds and layering them with real-world human context: the bizarre voice reviews, the Reddit and YouTube UGC, the Semrush search trends, filtering out the noise, and giving the AI exactly what it craves.

Leo  19:02

We’ve also explored the critical importance of an Omni LLM strategy, understanding that building a single API connection is a trap, and why utilizing an enterprise protection vault is the only sustainable, legally compliant way to execute this at scale.

Ada  19:16

Which brings us back to you listening right now. Whether you are a brand marketer trying to hit your quarterly revenue numbers, a developer sweating over complex API integrations, or just a consumer wondering why shopping online suddenly feels so much more intuitive lately. The very mechanics of how we discover the world are being rewired beneath our feet.

Leo  19:34

They really are.

Ada  19:35

We are transitioning from a frustrating world of typing keywords to a seamless world of expressing intent,

Leo  19:41

and that profound shift in behavior raises a fascinating, almost philosophical question for the future of commerce.

Ada  19:47

Ooh, let’s hear it.

Leo  19:48

We’ve established that AI shopping agents are becoming incredibly perfect at matching our exact human needs to specific product specs based on ingesting all this use. User-generated content, right?

Ada  20:01

Yes.

Leo  20:01

They can parse the data, read the reviews, and handle the feature comparisons flawlessly.

Ada  20:06

The AI handles the rational logic better than a human ever could.

Leo  20:10

Precisely. So, if the AI handles all the rational decision making, will the future of brand marketing be less about explaining what a product does and entirely about how a brand makes us feel outside of the AI ecosystem.

Ada  20:23

Wow! If the AI is doing all the explaining and the feature matching, the brand just has to do the emotional connecting.

Leo  20:29

Exactly.

Ada  20:30

That is something to definitely mull over the next time you ask a chatbot to recommend your next purchase. Thanks for diving deep with us today.