Mukund Jha believes that building a local company for the Indian market requires the same amount of effort as building a global company.

Mukund Jha believes a lack of focus contributed to Dunzo's scaling challenges, as the company pursued multiple business models instead of concentrating on its successful Darkstore model.

At the time Mukund Jha worked there, Google's search division was resistant to implementing machine learning.

Mukund Jha's previous company, Dunzo, raised approximately half a billion dollars in funding.

India accounts for approximately 10% of Emergent's revenue.

Emergent's competitive strategy was to focus on building fully functional software with real backends and databases, differentiating from competitors who focused on frontend demos.

95% of the Emergent team is based in Bangalore, India.

Emergent's go-to-market strategy heavily utilized influencers to launch and promote its product.

A four-person team at Emergent achieved the world number one rank on the SweepBench benchmark for coding agents.

Emergent was rejected by 10-12 VCs during fundraising because its pitch to fully automate software engineering was considered too ambitious.

At its peak, Dunzo was processing about 10 million orders per month.

Emergent's team predicted that foundational models would improve at generating JSON output, leading them to skip building a custom solution for JSON parsing.

More than 10 million applications have been built using the Emergent platform.

The current version of the Emergent product was launched approximately nine months before the interview.

Emergent has surpassed $100 million in annualized revenue run rate (ARR).

The majority of Emergent's revenue comes from the US and Europe.

The Emergent platform has more than 8.5 million users.

Emergent has rewritten its core system three times in nine months to adapt to the capabilities of new AI models.

Emergent's platform is built on a multi-agent orchestrated system with specialized agents for tasks like design and automated testing.

Emergent has developed a proprietary memory system that allows its AI agents to self-learn and improve from every new application built on the platform.