Indian startup Runable has raised $21 million to extend its AI agent from building websites and applications into finding customers and running growth work for small businesses. Susquehanna Venture Capital and Nexus Venture Partners co-led the Series A, with Together Fund and Array VC also participating.
The primary, all-equity financing values Runable at $65 million after the investment, according to co-founder and CEO Umesh Kumar. The Bengaluru company was founded in 2025 and employs 15 people.
A pivot from browser infrastructure to business outcomes
Runable originally developed browser technology for collecting data at scale. Its founders noticed that users increasingly asked the browser agent to create websites and presentations, prompting a move toward a general-purpose system that can build and deploy digital products from natural-language instructions.
The startup now wants to continue beyond creation. Planned capabilities include running advertising campaigns, managing social media, improving search visibility and optimizing how a business appears in AI chatbot results. Kumar frames the proposition around outcomes: a small company wants customers, not a collection of coding and marketing tools.
Runable says it reached a $2 million annualized revenue run rate within three weeks of launching payments in March. It reports 1.7 million registered users and more than one trillion tokens consumed over the past 90 days, with 60 to 70 percent of usage coming from paying customers. These are founder-supplied figures, and Kumar declined to disclose current revenue or the number of paying accounts.
Growth automation still stops at account boundaries
TechCrunch tested Runable by asking it to create and deploy a website for a fictional coffee subscription company, add analytics and attract 100 visitors with a $25 advertising budget. The agent built the site and prepared a campaign but could not run the ads until an external advertising account was connected.
A similar test with Cursor also reached account and payment barriers, while Runable handled more of the deployment infrastructure inside its own product. Runable said it can place ChatGPT ads without a customer’s own account through unidentified partners, but it declined to name those relationships.
The result illustrates the gap between preparing work and completing it. An agent can generate assets and configuration, yet spending money or acting through third-party services requires credentials, permissions and accountability. Runable’s promise of customer acquisition depends on crossing those boundaries safely and transparently.
Fast usage growth carries weak unit economics
Kumar acknowledged that Runable currently has negative gross margins because it subsidizes AI usage. The company combines external models with systems it is developing itself and expects falling inference costs to improve economics. He predicts comparable inference quality could eventually cost almost ten times less, but that remains a projection.
Runable also depends on model providers that are building their own agents. Kumar says coding-focused users working with local files may be better served by Codex or Claude Code, while Runable targets nontechnical owners who do not want to assemble deployment, analytics and distribution services. He names Manus and Genspark as closer general-agent competitors.
The largest markets currently include the United States, United Kingdom and Japan, with Japan expected by the company to become one of its top markets. Brazil also contributes users.
The financing gives Runable time to integrate more of the growth workflow, but its disclosed negative margins show the cost of providing a broad agent before the economics are settled. The next evidence is not another generated website. It is whether the system can acquire customers with permission, measure the result and charge enough to cover the intelligence and infrastructure behind the work.
Until those results appear, rapid adoption and sustainable demand remain separate questions.
