Back to Flash News

Replit Acquires Icelandic Duo Company Test13 to Use AI for Rapid Prototyping of Profitable SaaS, Keeping What Sticks

The Icelandic duo company Test 13 has joined Replit. Co-founder Viktor Margeirsson stated that eighteen months ago, he and Gunnar Kolbeinsson started with two beliefs: to use AI to rapidly prototype profitable SaaS, keeping what sticks; and to reduce local software service pricing to 20-40% of competitors. The platform made this possible.

He noted that in eighteen months, revenue exceeded $700,000, with a year-on-year growth rate of over 400%. The company rule is not to hire more people but to use agents to supplement the workforce, and they do not accept external financing. Viktor stepped down from his product manager position, claiming he cannot code, and after trying V0, Lovable, Base44, and Bolt, he got stuck at launch. His partner had used Replit in the past, and all their products have since been delivered on that platform.

Among their seven self-developed products, four were killed early. SMS marketing Diskó generated about $20,000 to $30,000 per month with a gross margin of about 70%; the OEM business had single orders ranging from $20,000 to $180,000. For the full year of 2025, they estimate around $120,000, enough to support the two founders; entering 2026, they conservatively estimated about $1.1 million with Diskó included, raising per capita revenue from about $60,000 to about $550,000. They also created Autobot, which they claim can cut custom software costs by 90%, delivering a runnable initial version in one hour, with no charge if unsatisfied.

Replit first selected them into its global ambassador program, with only ten spots available, and they are the only ones from Europe, based on their public platform demonstrating real deliveries. Viktor wrote to founder Amjad Masad, stating that the Icelandic market is too small and he feels like an observer on the frontier, wanting to place the next decade at the center of AI reconstruction. Months later, the deal was made. He has been included in the twenty-person AI advisory committee by the president's office and has led four founders in Replit's revenue competition.

The buyer's valuation is about $9 billion after a round of financing in March at around $4 billion, claiming to target a recurring revenue of $1 billion by the end of the year, with over 60 million global users, and has opened its first overseas office in London. The acquisition amount has not been disclosed. What was acquired is not just another small team but a set of methods already validated in a small market: fewer people, low prices, rapid kills, and using the same set of agents to parallel OEM and self-development.

In market mechanics, the buyer is a platform that needs cases to prove that "non-engineers can also create chargeable software," while the seller is a two-person studio that has turned the platform into a factory. This is a case-driven model recovery: ambassador content first helps Replit capture conversions from tools like Lovable, then incorporates the successful pricing and delivery into its system. Funds flow from the $9 billion valuation financing account to the acquisition price and the two salaries; the beneficiary is Replit, which needs a European benchmark, while the pressured parties are local software companies still pricing by person-days and investors who must explain why a $9 billion platform would buy a $700,000 revenue shop.

Source: Public Information

ABAB AI Insight

Replit has evolved from a browser-based coding classroom to a platform generating full-stack applications, with Amjad Masad increasing its valuation from about $3 billion to around $9 billion through the Agent series. Test 13's path is the opposite: first using Replit as a factory, then writing the factory's user manual as a customer acquisition tool. Viktor got fifteen out of sixteen competitors to try Replit in nineteen meetings, which is a cheaper channel than advertising. The acquisition turns the best teaching users into internal sales engineers, rather than just buying Diskó, the cash cow.

The capital logic is internalizing the model. The $9 billion platform does not need another SMS tool; it needs a sales-reproducible sentence of "two people, no financing, $700,000 in eighteen months, pricing down to 20-40%." Iceland is small enough for nationwide trial and error, and EEA rules make deliveries resemble European projects. Moving these two from Reykjavik to Replitopolis is equivalent to relocating high-density experiments from the geographic periphery to the valuation center, to counter the homogenized narratives of Lovable, Bolt, and Figma Make.

Analogies include GitHub acquiring the studio that excels at teaching open source and Salesforce buying the consulting firm that maximizes its own CRM. The industry is transitioning from "tools over functions" to "tools over who can sustain themselves with it" in the mid-expansion phase. Autobot packages agents as 10% custom, aiming for the platform to directly absorb the person-day bills of small and medium software companies, rather than just selling subscriptions.

Structural judgment belongs to the technological substitution driving the reconstruction of the industry chain. The mechanism is: the pricing power of coding wages shifts from local person-days to agent hours, with agents at about $10 per hour tearing apart quotes with five times the premium. The platform first subsidizes benchmark companies, then buys them back as manuals. The next round to be restructured is not the models, but the software workshops selling people by the hour.

ABAB News · Cognitive Laws

  1. The most expensive function of the platform is the users who have already made money for it.

  2. Companies that do not hire prove that tools have truly replaced staffing.

  3. Those who first drive prices down to 20% will force competitors to reprice.