Emergent: How Six Months of Tinkering Led To A $100M ARR Company

Predictions and Expectations:

• Mukund expects AI progress to be exponential and anticipates that coding is a space that will be disrupted very quickly.
• Mukund believes that if models are projected out a little bit, they will be able to solve steps they are currently failing at.
• Mukund predicts that the next model class will be able to solve JSON output problems, leading the company to skip addressing that issue immediately.
• Mukund expects that every time a new class of model comes out, the system must be reimagined from the lens of that new model.
• Mukund believes that thinking big and ambitious, specifically 10x or 100x current ideas, increases the probability of success.
• Mukund expects to see more AI-native companies coming out of India over the next year or so.

Timelines and Milestones:

• The current version of the project was launched approximately nine months ago.
• The team rewrote their system three times within nine months.
• Mukund states they will start rethinking their architecture where a new model is going to be in six months.
• Mukund expects to see more people from India in Emergent in the next year or so.

Technology and Product Direction:

• The company plans to build autonomous agents that can do more than just co-pilots, anticipating a future where parallel agents swarm together to complete tasks.
• The company intends to always build in the direction of AI, rather than solving problems piece by piece.
• The platform plans to continuously extract learnable aspects from new apps built on Emergent to make the platform self-improve via a large memory system.
• The company plans to automate all of software engineering, taking a massive view to build the best coding agent.
• The company plans to continue inventing infrastructure themselves, such as disk and memory snapshotting, to preserve state for parallel agents.

Market and Industry Outlook:

• Mukund expects that users want real software that works, noting that current solutions are good at starting but bad at finishing.
• Mukund observes that the latent demand in the market is very high for people who want to build software but lack access to tools.
• Mukund believes that building a global company is equally hard as building a local company, but the reach allows for global customer access from day zero.
• Mukund predicts that the tech landscape is changing such that it is a time to "attack the ceiling" rather than the floor.

Company Plans:

• The company plans to continue hiring and invites interested individuals to apply for roles.
• The company plans to grow by converting growth into a math problem to determine the necessary number of social views, impressions, and clicks.
• The company plans to use influencers as a primary strategy to launch the product to as many users as possible.
• The company plans to build a global company from India, mirroring the trajectory of Google and Facebook.

Financial Guidance:

• The company has recently crossed $100 million in annual run rate.
• The company is operating with a revenue run rate well over $100 million.

Risks and Caveats:

• Mukund is not sure if people know that before starting Emergent, he founded another company called Dunzo.
• Mukund notes that they faced rejection from most VCs who believed the technology was not there yet and that the idea was crazy.
• Mukund admits that initially, they were not sure how to coordinate an engineering team remotely while working from India.

Confidence and Disagreement:

• Mukund is confident that the models are capable of solving the problems they target, even if they are not currently perfect.
• Mukund is not sure that people realize the depth of tactical foundations the company has built compared to others.
• Mukund is certain that the expectation users have is for software that actually works, contrasting with other platforms that only deliver demos.