The Conversation Is the Interface - Building for the Post-Advertising Web

My name is Dee Taylor, and I'm the founder of Leyware, a startup building AI-native discovery and distribution infrastructure for video games.

Over the last month Leyware has performed 18,000 searches and after reviewing the data the only conclusion I am left with is that the advertising industry is poised to be seriously disrupted by conversational discovery. In this article, I'll walk through my results and what they mean for AI, startups, and advertisers in the years ahead.

How Leyware Search Works

Leyware Search is a conversational discovery engine for video games. Users search for specific games and we return results using a dual text and visual interface. I currently index 120,000 games across Steam and will scale to millions across all available platforms. The returned results are not ads, they are individual Steam pages adapted into a vector database that uses semantic search to match games to user intent.

I have analytic event flags assigned to when a user submits a search query and when a user clicks on 'View on Steam'. This button is in key locations of the user flow: directly available after each search on individual game cards and in the header for each expanded view. The user does not see impressions until after they perform a semantic search.

This enables me to understand how often people search and after receiving results how frequently do they visit individual Steam pages.

Results and Benchmarks

Leyware CTR vs paid UA for gaming benchmarks

• Leyware: 1.53% CTR, $0 CPM
• Facebook: 0.5%–1.6% CTR, ~$7–$8 CPM
• Google Search: 3%–4% CTR, ~$30 CPM
• YouTube: 0.5%–1% CTR, ~$3–$6 CPM
• TikTok: 1%–3% CTR, ~$5–$7 CPM

1.53% CTR with 4.48% user conversion rate exceeding typical benchmarks.

CPM might not be a fully applicable term post-AI since this is 'pull' vs 'push', but helpful for understanding how much developers are currently spending on advertising in existing channels. Game developers don't need to pay for conversational discovery, it simply exists.

Programmable CTR

Conversational discovery has superior or equal performance to paid advertising at zero cost. Conversational discovery is structurally more efficient because it is pre-qualified users that are intent driven. With an AI-native channel, traditional advertising becomes optional.

Now, these may sound like extraordinary claims and they are, but I am extremely confident in them because I am measuring these outcomes against real user behavior. My methods have not been optimized, do not use memory by default, and barely qualify as agentic. It's a prompt, a model, and an API call. The CTR I have been able to achieve is the floor for this application and will improve through additional context engineering, evals, observability, prompt optimization, and advanced agentic search methods.

Rather than using AI applications to optimize ad creative, ad copy, or ad strategy we can use software to optimize the CTR of AI powered product discovery and thus make it scaleable. As more users search, I receive more data that can be used to improve search results that will then help me acquire more users. This can be repeated across any consumer product (with proper LLM-ready infrastructure) and directs observable commerce traffic with zero marketing spend.

The Consequence of Hype

Unfortunately the first wave of AI marketing has fallen into the trap of optimizing a broken system rather than creating a new one that is AI-native:

AI Creative, AI Copy, and AI UGC - Generating infinite ad creative and ad copy lowers production costs but doesn't change the underlying channel costs. As the supply of ads increases exponentially, more ads compete for the same amount of attention, driving CPMs higher.

GEO/AEO - Stuck in traditional SEO thinking that believes search results are something that needs to be gamed or distorted. Especially since the 'optimization' component is to just churn out infinite website variations that contribute to the dead internet.

We are a mouse click away from near-AGI levels of intelligence and so far the primary innovations in marketing are to just automate the noise. It feels like we've invented the printing press just to make junk mail. I believe it is vitally important to think beyond existing frameworks and replace broken systems rather than optimizing them.

Imagine if ChatGPT, Gemini, and Claude had the ability to know and surface every product on the planet.

What would we need ads for?

What This Means for Developers

The implications extend far beyond gaming:

Zero Acquisition Cost - When users come to you with intent, you don't pay to interrupt them. Conversational discovery is pull, not push.

Compound Learning - Every search improves the system. More data = better recommendations = more users = more data.

Infrastructure Over Ads - Instead of buying attention, build systems that capture intent. The ROI compounds over time rather than resetting with each campaign.

First-Mover Advantage - The competitive landscape is still forming. Early presence builds reputation before the space becomes crowded.

The Future of Discovery

We're witnessing the early stages of a fundamental shift in how people find products. Traditional advertising optimizes for interruption—showing people things they didn't ask for. Conversational discovery optimizes for intent—surfacing exactly what someone described.

The $1 trillion advertising industry exists because discovery was broken. Search engines ranked by keywords and backlinks. Social feeds sorted by engagement. App stores listed by downloads and ratings. None of these systems actually understood what people wanted—they just approximated it through proxy signals.

AI changes that. When you can literally describe what you're looking for and get it, the entire edifice of paid discovery starts to look like legacy infrastructure.

Where We Go From Here

At Leyware, we're building the infrastructure layer for conversational commerce. Game discovery is just the beginning—the same principles apply to any consumer product category.

The companies that win in this transition won't be the ones with the biggest ad budgets. They'll be the ones who built LLM-ready infrastructure early, captured intent data, and optimized for conversational discovery before it became the standard.

Join the Leyware AI Waitlist and help shape the future of AI-native discovery.