EN
Back to the archive

The archive · AI & Models · Strategic decision · 2025–2026

Rocket.new bets vibe-coding on production apps: $15M seed, $4.5M ARR in 3 months

DhiWise founders pivoted to Rocket.new in 2025, betting production-ready AI-built apps beat prototype rivals; a $15M seed followed $4.5M ARR.

Rocket.new

The betThat vibe coding wins on 'day two' — production-ready apps, not prototypes — letting it beat Lovable and Bolt at the enterprise end.Scaling

What the business is

An AI app-building platform where natural-language prompts generate full production-ready applications — UI, backend, integrations, scaling — sold as monthly token-based subscriptions.

Starting capital$15M all-equity seed led by Salesforce Ventures, with Accel and Together Fund (September 2025).

How it started

Surat-based founders Vishal Virani, Rahul Shingala and Deepak Dhanak had built DhiWise, a design-to-code developer tool. As AI codegen took off, they saw users asking for outcomes, not snippets, and pivoted: 'We've moved from design-to-code utilities to an AI-driven build system.' Rocket.new launched in beta in June 2025 and within three months passed 400,000 users in 180 countries, 10,000+ paid subscribers and $4.5M annual recurring revenue.

What happened

In September 2025 the company raised a $15M all-equity seed led by Salesforce Ventures, joined by Accel and Together Fund, to sharpen go-to-market, open a Palo Alto headquarters and invest in proprietary models. In April 2026 it launched Rocket 1.0, adding Solve (business strategy) and Intelligence (competitive research) phases beside Build — a 'vibe solutioning' platform aimed at enterprises, not just hobbyists; Rocket.new's own materials later cited 1.5M users across 180 countries.

How it ended up

Still live and expanding: Rocket 1.0 extended the platform beyond app generation into strategy and competitive intelligence for enterprises, run from dual HQs in Surat and Palo Alto, with plans to double India's engineering and product staff. Whether it hit its $60–70M ARR target for June 2026 is not publicly documented in verified sources.

Background

Rocket.new is a Surat, India-based AI app-building platform founded by Vishal Virani, Rahul Shingala and Deepak Dhanak as the AI-first pivot of their earlier developer tool DhiWise. Launched in beta in June 2025, it turns natural-language prompts into complete, production-ready applications — UI, backend, integrations and scaling — rather than the quick prototypes produced by rivals like Lovable, Bolt and Cursor.

The bet was on 'day two': Virani argued the real market gap was between AI codegen magic and production reality. Rocket.new's first app took about 25 minutes versus ~3 minutes for competitors, but shipped with all essential modules; token-based pricing from $25/month filtered hobbyists while keeping a 50–55% gross margin. Within three months it counted 400,000 users in 180 countries, 10,000+ paid subscribers and $4.5M ARR.

In September 2025 Salesforce Ventures led a $15M all-equity seed joined by Accel and Together Fund, funding a Palo Alto HQ and proprietary model work. In April 2026 Rocket 1.0 added Solve (business strategy) and Intelligence (competitive research) to Build, positioning the company as a 'vibe solutioning' platform for enterprises. The founders' stated targets were $20–25M ARR by end-2025 and $60–70M by June 2026 — targets verified sources have not confirmed.

What has to be true

  • Production readiness, not speed, was the wedge: complete apps with backends and integrations justified slower generation and defended against prototype-fast rivals.
  • Token-based pricing from $25/month kept gross margin at 50–55% and filtered out hobbyists whose unit economics would be negative.
  • Pivoting an existing team and dataset (DhiWise) gave Rocket.new proprietary training data and a faster start than greenfield competitors.
  • Expanding from Build to Solve and Intelligence turned a coding tool into a broader 'vibe solutioning' bet on replacing product-manager-style work.

What can be applied

In a race optimised for prototype speed, slower but production-ready output became the differentiator; token pricing aligned users with cost, and the pivot to 'day two' opened an enterprise wedge.

Aftermath

As of September 2026 Rocket.new is live and scaling: Rocket 1.0 (April 2026) serves enterprises and SMBs with strategy, app-building and competitive intelligence from dual HQs in Surat and Palo Alto, and the founders report over 1.5 million users. The company has not publicly confirmed hitting its aggressive ARR targets, and its enterprise go-to-market is still being built out with the seed capital. No shutdown, acquisition or funding distress is documented.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases