Perplexity bets local AI can remake the business stack

Perplexity CEO: SMBs Could Drive the Next Economic Boom

Aravind Srinivas lays out a vision in which agentic software, local computing, and sovereign deployments give smaller businesses more control and leverage.

3 key takeaways

1Perplexity is expanding from answer engine into an agentic platform spanning search, browsers, research, and autonomous workflows.
2Srinivas argues hybrid and local AI can improve privacy, cybersecurity, and control as models become more capable.
3He sees smaller, more profitable businesses and durable workflow value—not raw user counts—as drivers of the next economic boom.

Don't miss

Srinivas warns that multi-agent systems can produce dangerous behaviors missed by evaluations focused only on a single main agent.

The brief

Perplexity CEO Aravind Srinivas describes a company moving beyond answers toward an agentic computer that combines search, browsers, research, personal context, and synthesis.

His product principle is simplicity: users should not choose models or tools manually, because the system should select the right machinery for each workflow.

The infrastructure bet is hybrid. Local hardware can protect privacy and sensitive data, while cloud systems provide frontier capability and scale when needed.

Srinivas extends that logic to sovereign deployments, imagining Perplexity systems running on customer-owned hardware under the control of governments, companies, or individuals.

The sharpest warning arrives with multi-agent systems: testing a main agent alone can miss dangerous behavior that emerges when sub-agents interact.

His broader economic thesis favors depth over reach: AI products create value through recursive improvements to workflows, potentially empowering smaller businesses without traditional network effects.