Rocket.new raises $15M from Accel, Salesforce Ventures

Rocket.new, a Surat-based “vibe-coding” startup that turns natural language prompts into production-ready code, has raised $15 million in seed funding led by Salesforce Ventures and Accel, along with participation from Together Fund. The round will be used to hire, for model R&D, and to grow go-to-market efforts as the company does battle with fast-growing rivals like Lovable, Cursor, Bolt, and Replit.

Production-grade vibe coding push aimed at real apps

Vibe coding — the responsive design of apps that feels less like writing code and more like describing intent — has been ascendant riding on the back of modern large language models. Rocket.new’s pitch: rather than spitting out super-short demo sites, its agentic system builds full-stack applications that are meant to be taken through deployment and scale, not just strut around. Vishal Virani, Rahul Shingala, and Deepak Dhanak founded the company after pivoting away from their previous startup providing developer tooling, DhiWise.

The startup combines frontier models from Anthropic, OpenAI, and Google’s Gemini with in-house deep learning systems built upon DhiWise data sets. In addition to code generation, the roadmap adds “agentic” workflows that can take on competitive research, product scaffolding, and iterative updates from chat-like prompts — all of which automate steps that so often slow product teams down.

Prioritizing development velocity, not just quick demos

Rocket.new (version 0.3) errs on the side of completeness at the expense of speed. The first build, the company claims, typically takes around 25 minutes to land — slower than tools that kick out a page in minutes — because it uses one pass to build navigation, auth, data models, and deployment plumbing. For many pros, that tradeoff is worth it if the net result is cutting days off wiring and refactoring.

Traction by the numbers: users, ARR, and growth targets

After a June beta release, Rocket.new reports 400,000 users in over 180 countries with over 10,000 paying subscribers and $4.5 million in annual recurring revenue. The company is aiming for $20–25 million ARR by year’s end and $60–70 million ARR by mid-next year, citing strong word-of-mouth and social virality as early propellants.

Usage tends to be more for “serious” builds than throwaway prototypes. ~12% of projects are e-commerce apps (grocery, clothing), ~10% fintech, 5–6% B2B tools, and 4–5% mental-health apps. Approximately 45% of its customers build native mobile apps, while 55% ship websites. Developers will often create a Supabase backend or use an existing one, and then trigger Rocket.new to build a native mobile client — a scenario where the stack integrates with existing ones rather than displacing them.

The customer base tends toward funnel-qualified builders at organizations like Meta, PayPal, KPMG, PwC, and Times Internet to date (mostly for side projects or quick internal tooling) — an encouraging sign of future enterprise pull.

How the Rocket.new platform works from prompt to deploy

Users write the product, target audience, data model, and integrations in natural language. Rocket.new then creates a repo with frontend, backend, auth, and deployment scripts — it can keep reacting through conversational edits. Bankosky says the company’s agentic approach will start taking on tasks like product research, variant testing, and roadmap-like work — removing some of that overhead currently taken care of by product managers and analysts.

Pricing, token limits, unit economics, and margin goals

Rocket.new offers a free tier limited to 1 million tokens, but pricing then begins at $25 per month for up to 5 million tokens. This metered approach keeps the product aimed at committed builders and SMBs, while maintaining gross margins that Computer Vision estimates to be between 50% and 55% currently; it now has a goal of between 60% and 70%, as it seeks to optimize inference costs and use proprietary models. Tokenized pricing also provides teams with more predictability around costs — a particularly significant factor as AI-driven development stretches beyond the realm of prototypes.

Go-to-market expansion plans and the competitive landscape

The United States contributes 26% of the revenue, with Europe contributing 15–20% and India 10%. In order to better serve its customers and enterprise prospects in the U.S., Rocket.new is opening a headquarters in Palo Alto and keeping its core engineering team in Surat. The company intends to double engineering and product headcount over the next year.

Competition is fierce. Lovable created an uproar around lightning-quick web app generation; Cursor integrates AI deeply into the coding process; both Bolt and Replit emphasize immediate scaffolding in the browser. Rocket.new is betting that production-quality outputs, mobile-first support, and agentic product workflows will help it stand out in a crowded category. Salesforce Ventures’ support might also help open doors with enterprise buyers, while Accel’s experience with developer platforms means it can offer operational expertise.

Industry research from firms such as McKinsey and Gartner also indicates that AI-assisted software creation is going from experiments to core workflows, but enterprise adoption depends on trust: code provenance, licensing, data privacy, and security labeling and guardrails. That’s where Rocket.new’s “build for production” positioning — and its emphasis on model tuning and governance as well — may be the most important.

What to watch next: adoption signals and execution risks

Key signals to watch here will be growth rates of ARR in relation to token costs, penetration into enterprise accounts, and the further development of the effects of Rocket.new on code generation. If the startup can keep build times compressed while still delivering quality, it could have carved a niche in vibe coding: less “demo magic,” more rolling-deployable software.

For India’s startup map, Rocket.new’s ascension from Surat reflects a broader change: AI product companies no longer have to be based in legacy tech hubs in order to win worldwide. The company raised new capital from Salesforce Ventures, along with Accel and Together Fund, as it now looks to do the work required of startups after they generate noteworthy early traction but before they lock in durable enterprise-caliber traction.