# The AI Scaffolding Layer is Collapsing. LlamaIndex's CEO Explains What Survives
The infrastructure layer developers built to ship LLM applications is collapsing. That includes indexing layers, query engines, retrieval pipelines, and agent orchestration systems that once felt essential. Jerry Liu, co-founder and CEO of LlamaIndex, argues this collapse is actually progress.
As models grow more capable, they need less scaffolding to function. Framework companies that built tools to help developers compose deterministic workflows face a shrinking addressable market. The deterministic layer is becoming less necessary.
Liu sees the real competitive advantage shifting elsewhere: context becomes the moat. Raw framework capability matters less than the data and retrieval strategies feeding into models. Companies that own high-quality, domain-specific context win. Those that just build generic orchestration tools lose.
LlamaIndex, originally positioned as an indexing and retrieval framework, must evolve. The company moves beyond shallow, deterministic workflows toward deeper integrations with proprietary data and context management. This reflects a broader industry shift where the abstraction layers of yesterday become commodity features inside stronger models today.
The startup that survives isn't the one selling frameworks. It's the one solving the context problem.