
The $9B startup that wants to create a billion new developers
Audio Summary
AI Summary
Amjad Msad, CEO and co-founder of Replit, discusses the company's mission to enable anyone who can read and write to build and deploy scalable applications without technical concerns. Replit, which recently raised $400 million at a $9 billion valuation, has evolved over 10 years, initially solving development and deployment environment complexities. In September 2024, Replit introduced "Vibe coding," abstracting away code entirely, allowing users to interface with AI using natural language. More recently, with Agent 4, they've added design interactions, enabling multimodal input like drag-and-drop on a canvas, with future plans for video and audio.
Replit uniquely bridges the gap between a dev tool and a consumer-facing product, not primarily marketing to traditional engineers. Msad, who started coding at a young age and built his first business at 13, found traditional developer tools cumbersome. He aimed to create joyful tools focused on creation rather than accidental complexity. While he initially sought to make programming accessible, Replit's updated mission is to create a billion new developers. They discovered that many developers enjoy the intricacies of setting up environments, akin to craftspeople building their own tools. Consequently, Replit's most engaged users are often tech-adjacent individuals, such as product managers, designers, and entrepreneurs, who have ideas but are bottlenecked by technical hurdles. In 2023, Replit explicitly shifted its focus to these "creators," whom Msad describes as a new generation of "AI-native developers." He likens this shift to the era of VB6, which democratized programming for many who couldn't code before.
Users are building a diverse range of applications on Replit, falling into three main categories: personal software, enterprise software, and entrepreneurial products. For example, a physical therapist and her husband, frustrated by offshoring development, built a sophisticated health tech app on Replit to track client progress with 3D body scans and range-of-motion tracking. Other examples include a SaaS solution for pool maintenance, software for sports clubs still using MS-DOS, and personal healthcare apps for managing rare conditions or tracking family chores. This demonstrates how domain experts can now build tailored solutions, improving various sectors of the economy.
In the enterprise, Replit addresses two primary use cases: product development and internal tools. Companies like Whoop have seen an order of magnitude increase in the number of ideas they can prototype and try, accelerating product development and the creation of new features and business lines. For internal tools and line-of-business applications, Replit enables departments like RevOps to build sales automations, quote configurators, and manage data flows without relying on engineering resources or creating new data silos with off-the-shelf SaaS tools, potentially saving hundreds of thousands or millions of dollars.
Replit's user acquisition strategy leverages a product-led growth (PLG) approach. People often discover the tool for personal use and then realize its potential for work. Msad notes a "neurological shift" that occurs when individuals understand they can solve problems with code, leading them to view the world differently. This organic adoption is then supported by a sales-assisted model for enterprises, where Replit's team educates and evangelizes, helping champions within companies demonstrate the value of the platform, often through hackathons. Replit also engages in top-down enterprise sales, often winning due to its established reputation for security and compliance.
Regarding the limits of Replit, Msad confidently states that entrepreneurs can build SaaS products, consumer apps, and automation tools on the platform. While it's not currently focused on building new cloud platforms or machine learning systems, its versatile nature allows technically knowledgeable users to build sophisticated things. For those relying solely on Vibe coding, numerous success stories demonstrate the platform's capabilities. Replit-native agencies are emerging, offering services that are 60-70% cheaper and more effective than traditional agencies.
Replit integrates with various services like Stripe, HubSpot, and Salesforce through a system of "skills" and APIs. As users interact with the Replit agent, it searches a database of these vetted skills and code snippets to provide the necessary functionalities, akin to Neo downloading skills in *The Matrix*. Building a community around Replit differs from traditional dev tools, as it requires more education and demonstration of possibilities. Replit's "DevRel" team acts more as educators, focusing on simplified documentation and video content. On the enterprise side, they often propose hackathons to demonstrate value, targeting entrepreneurial individuals within organizations who are resourceful and influential.
Msad reflects on Replit's experience with Y Combinator, highlighting the intense focus and rapid progress achieved in three months. YC instilled a mentality of setting ambitious goals and working tirelessly to achieve them, a practice Replit continues with its "agent releases" every four weeks. YC also taught the importance of compound growth, particularly the 7% week-over-week growth metric for bootstrapping new product lines. YC significantly expanded Replit's network, enabling them to secure early funding from prominent investors like A16Z, which Msad believes was crucial to their success.
Replit's Agent 4, their fourth major agent version, is aligned with the observed twice-yearly step changes in AI capabilities. It introduces parallel agents for asynchronous work, allowing users to design or plan other tasks while an agent is working in the background. A built-in canvas enables simultaneous design exploration. Agent 4 also facilitates teamwork, as users can fork new VMs and work in parallel, with the orchestrator managing task subdivision. A key feature is the ability to generate different contextual outputs from a single project, such as a mobile app from a web app, allowing users to run their entire company on Replit.
Looking ahead, Msad believes we are moving towards a "post-prompting world" where users give high-level goals to agents, such as "optimize my marketing funnel" or even "build me a SaaS company and make me some revenue." The necessary skills for this future include understanding what's possible, having a playful mindset, staying updated on AI advancements, and possessing persistence. Idea generation and creativity will remain crucial, as entrepreneurs will need to continuously generate new solutions.
If starting Replit today, Msad would prioritize culture, which they "screwed up" at one point, and be more honest about product-market fit. He emphasizes that true product-market fit is "an explosive thing" and easy to delude oneself about. He also expresses anticipation for improved "computer use models" in AI, which despite being seemingly easier to gather data for than language, have lagged. These models are crucial for testing user-built applications and providing nuanced UX feedback. Continual learning for agents, allowing them to improve on the job within an organization, is another highly anticipated development.
Msad envisions a future company composed primarily of "builders" and "salespeople." Sales roles will transform into evangelists and educators, helping companies transform with technology. Builders will focus on higher-level automation, with their roles continuously evolving as AI takes on more tasks. The future company will be driven by business generalists, essentially founders, who understand customer needs, the economy, and AI technology, using agents to solve problems across various departments. Replit already has an internal "Vibe coding in resident" team that identifies problems within the company, such as support queue prioritization or HR onboarding, and builds solutions using Replit, demonstrating this future model.