YC Root AccessEmergent: The AI App Builder for Everyone
CHAPTERS
- 0:00 – 0:18
AI as a ‘big reset’: why now is the moment to build boldly
Mukund frames AI as a once-in-a-generation platform shift, arguing we’re still at the very beginning of the AI era. He encourages builders to commit deeply to problems they care about and take bigger bets than they would in a more mature market.
- •AI will define the next ~20 years of technology and company building
- •We’re early—likened to “Bitcoin at dollar one” in terms of opportunity
- •Advice: pick a problem you love and go all-in
- •Sets the tone for Emergent’s mission: expanding who can build software
- 0:18 – 0:56
What Emergent is: prompt-to-production app building for non-coders
The founders explain Emergent as an AI app builder that turns plain-language prompts into production-ready applications. The product targets people with zero programming knowledge and supports shipping real apps, not just demos.
- •Users describe what they want; Emergent generates a launchable app
- •Designed for people without programming experience
- •Supports building mobile apps, web apps, and websites
- •Positioning: production-ready output rather than toy projects
- 0:56 – 1:35
Hypergrowth metrics: $15M ARR in 3 months, 1.7M users, 2.5M apps built
They share headline traction numbers shortly after launch and discuss deployment behavior on the platform. While many users deploy elsewhere, a significant portion can already deploy directly through Emergent.
- •$15M ARR achieved three months after launch
- •~1.7M users on the platform in that period
- •2.5M+ apps built to date
- •~20–30% of users deploy apps on Emergent today; many deploy off-platform
- 1:35 – 3:08
Founders’ backstories and the twin-brother operating dynamic
Mukund and Madhav describe their paths through PhD programs and major tech roles, plus Mukund’s experience building Dunzo at large scale. They also highlight how being twins creates high trust and fast collaboration.
- •Mukund: PhD dropout → Google → founded Dunzo (2,000 employees; large delivery fleet)
- •Madhav: CS PhD → Zenefits early employee → Amazon SageMaker team
- •They reunited post-Dunzo to start a new company together
- •Twin dynamic: high trust, constant idea-sharing, low coordination overhead
- 3:08 – 4:23
The YC journey: from testing agents to a general coding agent ambition
They recount entering YC with an initial idea around QA/testing agents that could validate apps via natural language. As agent capabilities improved, they realized the broader and more exciting challenge was a general coding agent capable of long-horizon tasks.
- •Initial concept: natural-language testing agents for web/mobile apps
- •Belief in autonomous agents over copilots for long-horizon work
- •Pivot insight: “general coding agent” is the bigger, more ambitious problem
- •Iterated through ideas while searching for the right wedge
- 4:23 – 5:35
Enterprise detour and deep agent R&D: SWE-bench and tight feedback loops
The team initially targeted enterprise use cases and invested heavily in agent research, achieving top performance on SWE-bench at the time. Madhav explains that agent autonomy requires tight feedback loops, motivating Emergent’s in-house infrastructure and multi-agent approach.
- •Focused on enterprise first; built a leading coding agent
- •Reached #1 on SWE-bench after intensive iteration
- •Key insight: autonomous agents need fast, tight feedback loops
- •Built infra in-house (backend/db/etc.) to enable agentic operation
- •Moved toward multi-agent techniques to improve reliability and quality
- 5:35 – 6:36
Why they switched to consumer: faster iteration and ‘idea-to-launched-app’ focus
They describe enterprise sales/feedback cycles as too slow and emphasize their natural consumer orientation from prior experience. Seeing internal success building apps with their system, they chose to go direct-to-consumer and leverage the strong foundations built during R&D.
- •Enterprise cycles slowed learning and product iteration
- •Founders identify as consumer-oriented builders (Dunzo background)
- •Internal usage proved the app-building workflow was compelling
- •Earlier R&D foundation became the differentiator at launch
- •Claim: users can feel quality differences vs other platforms
- 6:36 – 8:00
How Emergent builds apps: devbox environment, clarification chat, and routing
Mukund walks through what happens after a user hits enter: Emergent spins up a cloud dev environment so the agent can work like a real developer. The system first clarifies intent via conversation, then routes requests to the right agent(s) depending on what’s being built.
- •Spins up a cloud “dev box” (VM) for the agent to execute real dev tasks
- •Agent installs libraries, writes code, and gets automated feedback (e.g., linting)
- •Pre-build conversation to understand needs across non-technical users
- •Routes work based on intent (front-end vs mobile vs other requests)
- •Positioning: integrated support for web + mobile + backend in one platform
- 8:00 – 9:18
Differentiation: production-ready apps via integrated infra + multi-agent SDLC
They argue most vibe-coding tools stop at prototypes, while Emergent aims to take users from idea to launched, monetizable products. Their differentiation centers on integrating infrastructure with a state-of-the-art coding agent and orchestrating multiple specialized agents across the software lifecycle.
- •Goal: move beyond prototypes to fully launched apps
- •Integrated infra avoids reliance on third-party backend/db for core experience
- •Multi-agent architecture with specialists for design, testing, security, deploy
- •Deploy agent converts build steps into infrastructure-as-code and ships to their cluster
- •Unified experience enables building more complex apps end-to-end
- 9:18 – 10:00
Current limits and scaling complexity: pushing beyond typical code-size ceilings
They discuss practical ceilings for today’s AI-built apps and how Emergent compares to other tools on complexity. While very large codebases remain challenging, they expect the ceiling to rise with better models and platform improvements.
- •Typical Emergent apps: ~35k–40k lines of code (as a rough proxy)
- •They claim many competitors cap around ~10k LOC
- •Large apps (~100k–200k LOC) still strain current platforms/models
- •Expectation: ceiling increases as models and platform mature
- •Trajectory: from small prototypes → full-fledged apps → larger systems
- 10:00 – 12:18
Who builds with Emergent: real examples from scientists to small businesses to artists
The founders share diverse user stories showing how non-technical domain experts build bespoke software for personal and business needs. Examples include audiobook narration customization, jewelry repair pricing automation, and operations tooling for service businesses, plus filmmakers building striking sites.
- •Microbiologist built customizable audiobook narration workflow (e.g., ElevenLabs voices)
- •Jewelry owner built photo-based AI pricing engine for repairs
- •Gardener built a SaaS for job allocation, admin tools, and location tracking
- •Surprise: adoption extends far beyond PMs/designers to “all walks of life”
- •Software needs are personal; Emergent targets domain experts and small businesses
- 12:18 – 15:40
Launch and growth playbook: invite codes, influencer experiments, and data tracking
They explain a tightly controlled early rollout, starting with a tiny alpha and then an invite-only beta driven by influencers. By instrumenting invite codes, they measured conversion per influencer/content and learned what worked on TikTok, Twitter, and Instagram before opening access broadly.
- •Alpha testing with ~50–100 users to validate product quality
- •Recognized crowded market; needed standout distribution despite late entry
- •Influencer-led invite-only beta with trackable invite codes
- •Learned platform-specific content dynamics (TikTok/Twitter/Instagram)
- •Methodical measurement replaced “spray and pray”; later growth became more organic
- 15:40 – 17:13
Retention strategy: prioritize power users building real businesses (not ‘tourists’)
They address the ‘app tourist’ problem common in vibe-coding tools by focusing on serious builders with clear goals and longer prompts. These power users drive revenue and show very high retention, often replacing expensive dev-shop quotes with Emergent builds.
- •Many users try once and churn; Emergent optimizes for serious builders
- •Internal metric focus: “power users” (often signaled by longer prompts)
- •Power-user retention claimed at ~85–90%
- •Users often compare against dev-shop quotes (~$100k) and build for <$1k
- •Long-tail tourists may convert, but aren’t the core focus
- 17:13 – 18:54
Fundraising and scaling: $23M Series A, a 12-engineer team, and hiring plans
Mukund explains the Series A with Lightspeed closed rapidly after launch on early revenue signals and strong product impressions. They plan to invest primarily in building a top-tier team and continued agent research, while actively hiring across engineering, research, and marketing.
- •$23M Series A closed ~2 weeks after launch
- •Raised at roughly ~$2–3M ARR; investors tried the product directly
- •Team size at the time: ~12 engineers despite $15M ARR run-rate
- •Capital allocation: world-class hiring + research + platform improvements
- •Hiring focus: engineers, researchers (post-training), marketing/storytelling; new SF office
- 18:54 – 21:07
The ‘billion builders’ vision: custom agents, better quality, and a new startup wave
They outline near-term product expansion (mobile, custom agents) and a longer-term goal of perfecting the software engineering lifecycle experience. The conversation closes with a broader market prediction: AI tools like Emergent could unlock a massive increase in the number of people who can build and launch products.
- •Mobile app building exists already but is under-marketed; includes backend by default
- •Upcoming focus: custom agents for user-specific workflows
- •Core thesis: emulate what the best engineering team’s SDLC would look like
- •Market belief: “a billion builders” will emerge, driving a surge of new ideas/startups
- •Parting advice: AI is the reset—be bold and commit to problems you love
