Runable raised $21M and is betting on business "coming of age": the AI agent moves from launching projects to promoting them

20 September 202627 views

Bengaluru-based startup raised funding led by Susquehanna Venture Capital and Nexus Venture Partners at a $65 million valuation — the money will go toward making the agent not only build websites and apps on demand, but also bring in customers: running ads, managing social media, handling SEO, and optimizing for AI chatbot search results. Over the past three months, users have run more than a trillion tokens through the platform, and roughly two-thirds of that volume came from paying customers.

Runable raised $21M and is betting on business "coming of age": the AI agent moves from launching projects to promoting them

Deal: $21M, valuation $65M

Bengaluru-based startup Runable raised $21M in a Series A round. The round was co-led by Susquehanna Venture Capital and Nexus Venture Partners, with existing investors Together Fund and Array VC joining in. This is pure primary equity financing: there are no debt instruments in the deal, and the company's post-money valuation came to $65M. Co-founder and CEO Umesh Kumar shared the details in a conversation with TechCrunch (by Jagmeet Singh, August 26, 2026).

The company is very young: founded in 2025, headquartered in Bengaluru, with a team of just 15 people. The product is aimed at small businesses — people who have neither a development department, nor a marketing team, nor the time to figure out a website builder.

From browser scraper to universal agent

It didn't start with a landing page builder. Kumar and co-founder Saksham Sarda were building an infrastructure product — browser technology that could scrape data from the web at scale. The pivot happened naturally: the founders noticed that users were asking the browser agent for something entirely different — to put together a presentation, build a website, prepare content. That's how infrastructure development turned into a universal agent.

The economic effect was quick to show. Payments launched in March, and within three weeks Runable hit a $2M annual run rate. For a project that was selling technology rather than results just yesterday, that's a fairly sharp change in trajectory — and judging by the new round, investors believed it would hold.

From "build" to "promote"

Today the agent can create websites, apps, presentations, and other content from a natural-language request. It also takes on some of the basic infrastructure: deployment, analytics. But the main bet of the new round isn't launch — it's growth.

The company is expanding the agent toward the so-called "grow" side of the business: launching ad campaigns, running social media, SEO, and — most interestingly — optimizing a business's presence in AI chatbot answers. The idea is for a small business owner not to piece together a website, analytics, ad accounts, and marketing tools, but to simply ask for a certain number of customers.

Kumar's logic boils down to a simple question: if an entrepreneur pays an agency ten thousand dollars to manage Google ads, can't the same result be achieved for less? That's exactly the intersection where Runable wants to make money. He himself puts it more bluntly: a business owner doesn't need a tool for writing code, but a measurable result.

Audience and metrics

According to Kumar, Runable has around 1.7M registered users. Its largest markets are the US, the UK, and Japan, with the Japanese segment growing so fast that as soon as next month it could tie with the American one for the lead. There are users in Brazil too, but the company is deliberately shifting its focus toward the first three markets.

Here's what remains off the record: Kumar declined to disclose current revenue and the number of paying customers. As an indirect figure, he offered something else — over the past 90 days, users consumed more than 1 trillion tokens, and roughly 60–70% of that volume came from paying customers. That already says more about the product than the registration count: people aren't just trying the agent, they're running real tasks through it.

Economics: negative margin and a bet on cheap inference

That intensity of use costs money. Kumar openly admits: Runable's gross margin is currently negative, partly because the startup subsidizes AI usage for its customers. This is the classic story of growing agentic products — there's revenue, but it doesn't yet cover the inference bill.

The plan rests on two things. First: the company works with a set of third-party models while simultaneously developing its own to reduce dependence on others' pricing. Second: inference getting cheaper over time. Kumar estimates the potential for improving the economics at nearly a tenfold cost reduction while maintaining the same answer quality.

Competition: giants and coding platforms

The market Runable has entered is getting tighter. Model giants are moving in — Anthropic and OpenAI — while development platforms operate alongside: Cursor, Lovable, and Replit. Any of them could add agentic features, and then the advantage of niche players would start to erode.

The main challenge is essentially this: standing out against companies whose models you yourself use. Kumar's argument isn't the model, but the amount of work around it. Runable takes on infrastructure, analytics, and distribution so the user doesn't have to glue five services into one process. The model as such stops being the product — the completed task becomes the product.

Who Runable isn't competing with

Kumar isn't trying to beat coding agents at their own game — and he says so openly. A developer who works with local files and spends most of their time writing code would be better served by OpenAI Codex or Anthropic Claude Code. That's a different use case and a different user.

Runable's target audience is non-technical small business owners who care not about the process but about a working business at the end. Kumar names the universal agents Manus and Genspark as the closest competitors — that is, products in the same category, not developer tools.

TechCrunch's test: what the agent did and where it stumbled

The journalists put the agent through a practical test: a fictional coffee subscription business, with the task of building and deploying a website, setting up analytics, and bringing in the first 100 visitors on a $25 ad budget. Runable created the website and prepared an ad campaign, but didn't launch it: first, it said, an ad account needs to be connected.

A similar test with Cursor ran into some of the same limitations — the campaign was prepared, but access to an ad account and a payment method were required, and ongoing website hosting required a third-party service. The difference is that Runable closed more of the infrastructure inside its own platform. But an external ad account was still needed before the budget could be spent.

The company responded to this point: the ability to launch ads without connecting client accounts currently works for advertising in ChatGPT. The startup has partnerships that make this possible, but it declined to name the partners, describing the relationship as a "soft wedge" — a way to enter the task from the side where the barrier is lower.

What this means

A $21M round at a $65M valuation is a bet not on "yet another website builder," but on changing the entry point into the product. If small businesses used to buy tools and assemble a process from them, now Runable offers to sell the result: not a website, not a campaign, not a report, but customers.

The weak spot of this model is visible to the naked eye — the economics. A negative gross margin at 1 trillion tokens consumed means growth is still being paid for by investors, not customers. The second vulnerability is dependence on others' models and others' ad platforms, each of which has its own agentic ambitions. The winner will be whoever turns the subsidy into margin faster and proves they take a task all the way to the end, not just halfway.

Frequently asked questions

Related materials

All materials
Validating Candidates Costs More Than Generating Them: Honeypots Against LLM Vulnerability Hunting
News

Validating Candidates Costs More Than Generating Them: Honeypots Against LLM Vulnerability Hunting

The study examines the cost asymmetry in agentic vulnerability discovery: generating candidate versions is cheap, while confirming them through reachability analysis, execution, and PoC development consumes most of the budget, making verification a potential standalone target for defense. The proposed RedHerring—repository-adapted decoys with a hidden certificate—reduced the number of real vulnerabilities found by 38.7–60.4% at equal cost across 33 OSS-Fuzz projects; even after the agent was warned, the reduction remained at 37.2%.

9 November 2026
Checking LLM Hallucinations via SQL-Based Source Representation Without Fine-Tuning
News

Checking LLM Hallucinations via SQL-Based Source Representation Without Fine-Tuning

The study examines whether an LLM can apply a basic symbolic skill—building a SQL database from source documents—to ground its outputs in sources without supervision when detecting hallucinations. On RAGTruth and DiaHalu, this neuro-symbolic pipeline proves more accurate than direct prediction and comparable to state-of-the-art methods, without requiring domain-specific fine-tuning.

9 November 2026
MathAdv: What the Final Accuracy of Machine-Generated Proofs Conceals
News

MathAdv: What the Final Accuracy of Machine-Generated Proofs Conceals

A diagnostic suite of tasks spanning 13 undergraduate and graduate mathematics topics evaluates machine-proving systems in Lean 4 not only by their final results, but also by their subject-matter mastery, the quality of their informal reasoning, and their robustness to rephrased conditions. A component-by-component analysis shows that translating a problem into formal form remains the main obstacle, and that success depends heavily on the area of mathematics and how the problem is framed.

9 November 2026
Runable raised $21M and is betting on business "coming of age": the AI agent moves from launching projects to promoting them