Skyfall AI Hands Three Acquired Businesses to Its Steve-001 AI System

Skyfall AI is launching Steve-001, an autonomous business harness that already runs three small digital businesses it acquired. A human owner sets direction and approves major decisions. One business more than doubled monthly revenue, one slipped about 3% below its purchase level, and the third is too new to assess.

Can an AI system run a real business?

Skyfall AI is launching Steve-001, which the San Francisco startup calls an autonomous business harness. The system already runs three small digital businesses that Skyfall has acquired. A human owner sets the direction and signs off on the decisions that matter most. The early numbers, which Skyfall shared ahead of the launch, are mixed: one business has more than doubled its monthly revenue, another has slipped slightly below its level when Skyfall bought it, and the third is too new to assess.

The company grew out of a team with a track record in AI research. Sam Pasupalak and Kaheer Suleman co-founded Maluuba, the Waterloo-based deep learning lab that Microsoft bought in January 2017. A few months after that sale, a Maluuba team reached the maximum Ms. Pac-Man score of 999,990 by splitting the game among more than 150 cooperating agents. Skyfall, which Pasupalak and Suleman started with Pasupalak’s brother Sumit, tested its ideas on games before moving to real money.

Late last year, Skyfall published Mini Amusement Parks, a business simulator modeled on RollerCoaster Tycoon. Human players ran a park several times better than the leading large language model agents, according to the company.

From simulator to acquisitions

In July, Skyfall came out of stealth with a plan to spend up to $1 million on a small B2B SaaS or ecommerce business and double its revenue within six months. Forbes first reported the plan. “If you want a truly autonomous business, you have to buy a business and operate the whole thing end to end,” Pasupalak told Forbes at the time.

BetaKit reported that Skyfall was looking for software companies run by their owners or by a small group of contractors and staff. According to BetaKit, any workers would receive a “generous departure package.” Skyfall has not said whether anyone worked at the three businesses it went on to buy.

Steve, the network of AI agents that runs the businesses, has already had to manage its own internal problems. During a routine website update at one business, two software agents got stuck passing the same task back and forth. A growth agent assigned the deployment to an engineering agent, which sent it straight back. Neither agent took responsibility, and one lacked the access needed to make the change. Steve eventually changed its own rules to prevent a repeat. It now has a coordinating agent that steps in when agents disagree and decides how to resolve the issue.

Two businesses, two directions

Did the acquired businesses grow under AI control?

The clearest result comes from a social media API tool Skyfall took over in July. Its monthly revenue had been flat through the spring and stood at $3,649 at the time of the takeover. By October, it had reached $8,307, an increase of 127.7%.

Skyfall credits the gain to a series of small changes rather than one big bet. Steve started answering questions at signup, and Skyfall now puts signup-to-paid conversion at 14%. Steve also stopped promoting one-off credit packs, which the company says added about four new subscriptions a week. A third change targeted cancellations. Discounts to stay are now offered only to newer users, which Skyfall says saves $120 a month in recurring revenue that would otherwise have leaked away.

The second business, a Shopify referral app, is a harder case. It had lost revenue for five years, along with heavy churn and a stretch when no new customers were paying at all. Skyfall bought it in August at $8,634 a month. It now brings in $8,379, about 3% less. The company calls that stabilization. Its chart shows a September dip turning into a small rise in October, with new customers signing up again.

Most of Steve’s work on the app has focused on getting it discovered. The system set out to improve the app's LLM visibility, making it easier for AI assistants to recommend, rewrote the App Store listing to explain how referral sales work, and switched off ads that drew clicks without producing purchases. Skyfall says that last change saved more than $1,000. The third business joined Steve too recently to produce meaningful figures.

How the system checks itself

Skyfall describes the engine underneath as a code world model. Each business is represented as an executable program, built from its own records, that models how customers, product usage, revenue, and costs affect one another. Steve runs candidate moves through that model before trying them on real customers. Each proposal comes with a forecast, a timeframe, and a confidence level. Once a move plays out, the gap between prediction and result feeds into the next version of both the model and the plan.

When a forecast misses, the system traces the error to the part of its setup that caused it. It then rewrites its own code, skills, tests, or agent configuration, and keeps a record of each change and whether later runs improved. In one case, Steve tested variations of its lifecycle emails, kept the versions that performed best, and folded them into its own instructions.

How does Steve-001 keep its changes under control?

Letting software modify itself while running a company with paying customers raises an obvious question about control. Skyfall’s answer is a policy engine that sits in front of every action. It checks the action’s type, the money involved, and its effect on customers against the owner’s rules and any rules the system has learned. Every decision falls into one of four tiers: proceed automatically, proceed and alert the owner, wait for approval, or stop. A refund of $30 or less might go through on its own, while anything larger waits for a person. Steve’s changes to itself must stay within those limits.

Outside investors and regulators are working on the same problem. Asymptote Labs raised $5.5 million last month for tools that record and block risky agent actions. In October, the UK’s data regulator began asking AI developers how their agents behave.

Earlier attempts to hand a business to an AI have been humbling. In 2025, Anthropic let a version of Claude run a small shop in its San Francisco office. It sold tungsten cubes below cost and ended the first month at a loss. A later, upgraded version turned the shop profitable.

Skyfall’s experiment runs on real businesses whose customers can cancel. The company plans to make its work visible. Each business will get a public dashboard under an anonymous name, where readers can track its targets, metrics, agent activity, decisions, and the changes Steve makes to itself. Skyfall plans to take Steve into other kinds of business next. For now, the July target still stands: double a business within six months. The social media tool has already cleared that bar. The Shopify app has not come close.

Este artículo fue escrito con la asistencia de IA.
News Factory APP - noticias agénticas para impulsar tu SEO y AEO.