The first wave of AI helped businesses do what they already do, faster and cheaper. AI writing assistants, code copilots, customer service bots. They replace human effort with machine effort. Same output, lower cost. That's valuable. But the net impact is cost-cutting. It doesn't directly grow revenue.
The next wave is different. It's already happening. Language learning apps, AI companions, fitness coaches, creative tools. These aren't cheaper versions of existing products. They're new categories, experiences and products that couldn't have existed before AI. They create genuine value for consumers, the kind that motivates spending.
What separates first-wave from next-wave AI is simple. First-wave does existing things cheaper. Cost-cutting is the impact. Next-wave does new things that weren't possible before. Net new value is the impact.
Building next-wave applications requires getting three things right to reach consumers: they have to be realtime, scalable, and personal.
Realtime. Consumers don't wait. Two seconds of lag mid-interaction and users leave. They don't come back. First-wave AI can be async: submit a document, wait for results. Next-wave AI has to feel instant. This is essential to making AI feel human and accessible.
Scalable. These applications need to serve millions of users reliably, without bankrupting the company. Cost structures matter. If you can't afford to grow, you can't reach everyone who wants what you're building. This requires delivering best in class models efficiently and at orders of magnitude lower cost.
Personal. The best user outcomes require different AI configurations for different people and contexts. A teenager in Brazil learning English for fun needs a different experience than an executive in Tokyo learning it for work. Different models, different voices, different everything. And finding what works for these different users and contexts requires experimentation: controllable models you can adapt to users, outcomes-based routing that selects the right configurations to optimize retention, conversion, time spent. Consumer companies ten years ago got good at A/B testing buttons and flows. AI is just another layer to experiment across. The winners will be those who experiment fastest and automatically adapt AI to optimize for each user’s success.
The opportunity is here. AI-native apps generate twice the revenue per install of traditional apps. Companion apps went from $0.50 to $1.20 revenue per download in one year. 95% of AI spend is still B2B. The consumer side is wide open.
Are you building tools that cut costs, or experiences that create new value?