a16z just dropped their State of Consumer AI 2025 report. The headlines focus on the LLM wars—ChatGPT’s dominance, Gemini’s growth, Claude’s prosumer focus.

But buried in the analysis is a more important insight for anyone building or buying marketing technology:

The major AI labs are not coming for vertical applications.

They’re focused on models and core features. That leaves an enormous opportunity for startups building dedicated, opinionated experiences in specific domains.

Like marketing.

The Landscape in 2025

Here’s what the report shows about consumer AI usage:

Player	Position	Focus
ChatGPT	Dominant (800-900M WAU)	Adding features to existing interface
Gemini	Fast growth (155% YoY desktop)	Model improvements, NotebookLM experiments
Claude	Prosumer/technical niche	Professional features, Claude Code
Perplexity	Productivity hackers	Search, browser, commerce
Grok	Fastest capability gains	Companions, media generation
Meta AI	Platform integration	AI embedded in existing social apps

The pattern: every major player is focused on being the general-purpose assistant or improving their core model capabilities.

None of them are building vertical marketing solutions.

”Winner Take Most” — But Most of What?

The report describes the LLM assistant race as “winner take most, not winner take all.” ChatGPT leads, but there’s room for others.

But here’s the thing: they’re all competing for the same job—being your general-purpose AI assistant.

That job is:

Answer questions
Help with writing
Code assistance
General research
Creative tasks

What that job is not:

Understanding your specific brand
Developing marketing strategy
Creating content in your voice
Managing multi-channel campaigns
Optimizing based on your performance data

The general assistants are incredible at general tasks. They’re mediocre at specialized ones that require domain expertise and persistent context.

The Vertical Opportunity

The report’s key insight for startups:

“There is significant opportunity for startups to build dedicated, opinionated consumer experiences, as the major labs are primarily focused on the underlying models and enhancing their core product features.”

“Dedicated” means: built for one domain, not trying to be everything.

“Opinionated” means: having a point of view about how the job should be done, not just providing generic capability.

This is exactly the gap in marketing AI.

Why General Assistants Fail at Marketing

Try asking ChatGPT to run your marketing:

You: “Create a social media strategy for my B2B SaaS startup.”

ChatGPT: Gives you a generic framework about audience research, content pillars, posting frequency, engagement tactics.

That’s useful information. It’s also what you’d find in any marketing blog post from 2019.

What ChatGPT can’t do:

Know your brand voice without re-explaining it every session
Understand your competitive positioning
Remember what worked in your last campaign
Adapt based on your specific audience’s engagement patterns
Actually execute the strategy it recommends

The general assistant gives you information. A vertical marketing AI does the work.

What “Opinionated” Means for Marketing AI

Generic AI tools are flexible. You can use them for anything. That flexibility is also their weakness—they don’t have opinions about how marketing should be done.

An opinionated marketing AI says:

“Your brand voice is X, so this content should feel like Y”
“Based on your audience, this channel matters more than that one”
“This campaign element isn’t performing—here’s the adjustment”
“This content doesn’t match your brand guidelines”

Opinions require context. Context requires specialization. Specialization requires focus.

The major labs aren’t building this because it doesn’t fit their model. They want to be everything to everyone. Marketing AI needs to be one thing, deeply.

The Integration Gap

Another insight from the report: users overwhelmingly stick with one assistant.

“The vast majority of consumers using only one general assistant.”

This creates a problem for marketing. Effective marketing requires integration across:

Content creation
Social media management
Email campaigns
Analytics
Brand asset management
Campaign coordination

If you’re using ChatGPT as your general assistant and then switching to separate tools for each marketing function, you lose the thread. The AI that writes your content doesn’t know what performed well. The AI that analyzes performance doesn’t know your brand voice.

Vertical marketing AI solves this by being integrated across the marketing workflow—not a tool you switch to, but a system that handles the whole job.

What This Means for Marketing Teams

If you’re a marketer watching the AI assistant wars, the report suggests a few things:

Don’t wait for OpenAI to build marketing tools. They’re focused on being the best general assistant. Marketing-specific features aren’t their priority.

Expect vertical AI to get better faster than general AI at specialized tasks. A company focused entirely on marketing AI will ship more relevant features than a company trying to improve coding, writing, research, and a hundred other capabilities simultaneously.

The “prosumer” trend benefits marketing. Claude’s focus on professional users with features like Artifacts shows there’s demand for AI that goes beyond chat into actual work output. Marketing AI should follow this pattern.

Integration beats features. The report shows limited cross-usage between assistants. Users want one tool that does the job, not five tools they have to coordinate.

Our Read

The State of Consumer AI 2025 is bullish on general assistants and bearish on their ability to dominate every vertical.

For marketing, this means:

The big labs aren’t coming to compete directly
There’s space for dedicated, opinionated marketing AI
Integration and specialization beat raw model capability
The window for building vertical AI is open

The labs will keep making better models. That makes vertical AI better too—we build on their infrastructure. But the product layer, the marketing-specific intelligence, the brand understanding—that’s ours to build.

The full a16z report is worth reading. And if you’re looking for dedicated marketing AI rather than general-purpose assistants, that’s exactly what we’re building.