How Omneky is rewriting the future of advertising with Hikari Senju

Digital advertising has a dirty secret: most of the budget is wasted. Not because targeting is broken, but because the creative is. Consumers see the same ad four times and conversion rates drop 50%. Meta's algorithm now demands thousands of creative variations to drive real performance. And most marketing teams - even at large enterprises - are woefully understaffed on designers.

Hikari Senju saw this coming years before the rest of the industry. He founded Omneky in 2019, integrated GPT-2 embeddings before most people had heard of language models, and built what is now the leading AI-powered omnichannel ad generation and optimization platform. In this episode, Hikari walks through the mechanics of creative fatigue, why advertising is expanding from cost center to revenue center, and how a closed-loop AI system can generate and optimize thousands of ad variations with no human in the loop.

The Creative Volume Problem

The fundamental math of modern digital advertising has shifted. Meta's algorithm now rewards volume of creative - not just quality. If you're not running thousands of variations, you're not competing. And yet the industry's staffing model hasn't changed: designers are chronically underprovisioned even at the largest enterprises.

The core insight driving Omneky is simple but underappreciated: what's expensive in advertising is the impression - the cost to serve the ad - not the cost to create the creative. Every impression you waste showing a fatigued ad is money thrown away. AI can generate infinite creative variations. That changes the economics of the entire industry.

The research is clear: when a consumer sees the same ad more than four times, conversion rates fall by 50%. Creative fatigue isn't a brand concern - it's a direct P&L problem. More variations, more relevance, more impressions that count.

How Omneky Works: The Closed Loop

Omneky is not a creative tool. It's a closed-loop optimization system that happens to generate creative. The distinction matters.

The platform integrates directly with your ad platforms - Meta, Google, LinkedIn, TikTok, Snapchat, Pinterest, Reddit - and ingests live performance data. It learns which keywords, imagery, copy styles, and messaging angles are actually driving clicks and conversions. Those insights get injected into AI prompts. The AI generates new ad creative - images, video, copy, layout - that incorporates the winning elements. Those ads get launched back into the ad platforms. The loop closes automatically, continuously, 24/7.

If "man in a blazer" is outperforming other visual treatments, Omneky generates more ads featuring a man in a blazer. If the phrase "data driven" is driving outsized click-through rates, it goes into every prompt. There is no human selecting winners and briefing designers. The system learns, generates, and deploys autonomously.

Hikari reports that Omneky runs its own ads profitably through this system - spending $1 to acquire a customer and generating $2 back through subscriptions and upsells.

Advertising as Revenue Center, Not Cost Center

Historically, advertising has hovered around 1% of global GDP and been treated primarily as a brand-building cost center. Hikari's thesis is that personalization and attribution technology are changing that categorization fundamentally.

When you can generate thousands of variations, target them with precision, and close the loop on what's driving actual revenue - not just clicks - advertising stops being a spend line and starts functioning as a sales channel. A targeted, personalized video ad communicates a value proposition more richly than a cold email subject line. If the attribution proves the revenue, the budget moves from marketing to sales operations.

He frames the advertising market as genuinely expanding as a result - not just shifting share among existing players, but growing the total addressable role that ads play in the revenue generation stack of businesses.

Who Should Be Using This

Any company running digital ads can benefit. Omneky's customer base skews toward e-commerce, SaaS, FinTech, healthcare, and crypto - but the underlying problem (not enough creative variation to drive performance, no closed-loop optimization) exists across virtually every vertical running digital campaigns.

The product is fully self-service. Hikari notes that a new user can sign up, generate ads, and launch campaigns within two minutes. A two-week free trial requires no sales call.

The Regulatory Overhang

When asked about AI limitations, Hikari is direct: the technical capabilities are improving so fast that any specific shortcoming he names today may be resolved within a month. Google's video generation was dramatically better at recording time than it was 30 days prior. The goalposts move monthly.

The one area with a slower improvement curve: legal and regulatory clarity around AI-generated content. Questions about copyright ownership, liability, and the intellectual property status of AI-generated creative remain genuinely unresolved. Court rulings have raised questions about the copyrightability of some AI-generated work. This is the area where Hikari sees the most persistent friction - not because it's technically hard, but because the legal system moves on a different clock than the technology.

The Founder's Path: Harvard, MIT, Two Startups, and a GenAI Epiphany

Hikari studied computer science at Harvard with a focus on machine learning and AI. In high school, he built a clothing recommendation website - an early recommender system. A machine learning course at MIT introduced him to generative AI nearly a decade before it became mainstream. His father is an artist, and the intersection of AI creativity with his personal interests made generative models immediately compelling.

In college, he co-founded two venture-backed companies, including a personalized learning application that was acquired by a Bay Area company where Hikari then ran marketing. It was the combination of that hands-on marketing experience - understanding creative at scale, ad performance, and the bottlenecks of the traditional agency model - with his GenAI background that surfaced the Omneky opportunity.

Omneky launched in 2019-2020, integrating GPT-2 embeddings from Hugging Face for early customers before most of the industry knew those tools existed. The company integrated GPT-3, then GPT-4, then Stable Diffusion, then DALL-E, and continues to integrate cutting-edge generation models as they emerge. It has been an AI company from day one.

Attention Is the Scarce Resource

Hikari's framework for thinking about where entrepreneurial value will be captured in an AI-abundant world is built around scarcity. When content is abundant - images, video, apps, games, all generatable at near-zero cost - what becomes scarce? Attention. Human eyeballs are a fixed resource. Their number isn't growing exponentially.

That's why Omneky is an advertising company: it's competing in the arena where scarcity is permanent. As content floods every channel, the competition for each impression intensifies, which makes the tools to win those impressions more valuable, not less.

On the question of AI agents doing our browsing and purchasing: Hikari isn't worried. Humans will always seek entertainment. Any experience - digital or physical, human-facing or agent-mediated - that captures human attention creates an advertising opportunity. Advertising, he notes, is as old as civilization. The cave painting pointing to where the buffalo are was an ad.

Tools & Resources

• Omneky - AI omnichannel ad generation and optimization platform; closed-loop system integrating with Meta, Google, LinkedIn, TikTok, Snapchat, Pinterest, Reddit; free trial available (omneky.com)
• GPT-2 / GPT-3 / GPT-4 (OpenAI) - Integrated sequentially since Omneky's founding in 2019-2020; embeddings from Hugging Face were the first production AI integration
• Stable Diffusion - Integrated for image generation; part of Omneky's evolving AI model stack
• DALL-E (OpenAI) - Integrated for image and ad creative generation
• Hugging Face - Source of the GPT-2 embeddings Omneky deployed for early customers before widespread LLM access
• Acceleron - Hikari's recommended sci-fi book; explores an optimistic future of AI-human collaboration, Dyson spheres, and post-scarcity civilization

Key Frameworks from This Episode

Creative Fatigue Economics

Seeing the same ad more than four times drops conversion rates by 50%. The expensive part of advertising is serving the impression, not creating the creative. AI eliminates the constraint on creative volume, making every impression count rather than burning budget on fatigued ads showing to disengaged audiences.

The Closed-Loop Ad System

Omneky's architecture: ingest performance data from ad platforms → identify what keywords and imagery drive results → inject those signals into AI generation prompts → generate new creative at scale → launch and measure. The loop runs autonomously, continuously improving without human briefings or creative reviews.

Advertising as Revenue Center

Traditional advertising = cost center for brand building. Personalized, attribution-linked, AI-optimized advertising = revenue center that can replace or supplement sales functions. When you can prove that $1 in ads returns $2 in revenue, the budget category changes - and the market for advertising tools expands.

Scarcity as Startup Strategy

In a world of abundant AI-generated content, attention is the fixed resource. Smart founders identify what remains scarce when AI makes something abundant, then build around that scarcity. For Omneky, infinite content generation makes the competition for finite human attention more intense - and more valuable to win.

Personalization at Ad Scale

A good salesperson personalizes their pitch for every prospect. Traditional advertising sends one script to millions. AI-generated creative variation closes that gap - thousands of tailored messages for thousands of audience segments, each addressing specific pain points and interests. The infrastructure for personalization now exists; the constraint was always creative production volume.

Information Asymmetry as Founder Edge

Hikari's framework for sustainable entrepreneurship: the founder who sees an opportunity others miss needs an information asymmetry advantage - reading more broadly, thinking differently about risk, spotting the pattern before the consensus forms. Being early on GPT-2 in 2019 when nobody was watching was an information edge, not luck.

FAQ