The Easiest Software Business to Start in 2026

TLDR
Liam Ottley introduces OpenClaw/Claudebot as a new platform that significantly lowers the barrier to entry for starting a Software as a Service (SAS) business through "skills," outlining five categories from simple prompts to data-backed services. While this opportunity simplifies SAS development, he strongly cautions that it is unproven and rapidly shifting, likening it to "building on sand" for most beginners due to volatility and business readiness issues. Instead, Ottley advocates for the "rock-solid" path of an AI auditing and consulting agency, which he argues offers long-term stability by addressing the human and organizational challenges of AI adoption.

Summary
The video begins by highlighting the traditional appeal of Software as a Service (SAS) – recurring revenue and high valuation – but underscores the immense difficulty of starting one. Historically, building a SAS required extensive development in user interface, infrastructure, and marketing, making the actual valuable functionality a small fraction of the overall effort, demanding significant time and capital, and thus being largely inaccessible to most individuals.

Liam Ottley then introduces OpenClaw, also known as Claudebot, as a revolutionary personal AI assistant and a platform akin to the iPhone, complete with a marketplace called Clawhub for "skills." Skills are essentially instruction files that teach the AI assistant new capabilities, ranging from analyzing legal contracts to pulling YouTube transcripts. Crucially, some skills can connect to external software and services. This "skills as a SAS" model drastically simplifies the process of building a software business: developers only need to create the core functionality, as the AI assistant (or platforms like WhatsApp/Telegram) serves as the interface, Clawhub handles distribution and marketing, and API keys streamline user authentication. Ottley notes that AI itself can act as a co-pilot, assisting with everything from idea generation to coding and marketing, allowing even non-technical individuals to build something impactful in a weekend.

He categorizes five types of skill-based businesses. Firstly, "Pure Prompt Skills" involve packaging expertise into a text file of instructions, offering low defensibility but serving as a starting point. Secondly, "Utility Skills" wrap a small script to perform a specific function (e.g., a high-quality YouTube transcriber), with recurring revenue derived from ongoing maintenance. Thirdly, "API Integration Skills" teach the AI how to interact with existing tools like CRMs. The fourth category, "Backend Service Skills" (Skills as a SAS), involves running a proprietary service on a server that the AI skill connects to, offering strong defensibility and consistent monthly revenue as users pay for access to the hidden server's heavy lifting. Finally, "Proprietary Data Skills" (Moat Builders) are backed by unique, valuable data stored in a vector database, providing the highest defensibility and revenue potential. He also touches on hosting OpenClaw securely on a Virtual Private Server (VPS).

Despite the excitement, Ottley issues a strong caution, likening this new opportunity to "building on sand." He stresses that it is unproven, volatile, and constantly shifting, favoring "cracked AI engineers" who can adapt at lightning speed. Furthermore, he points out that businesses cannot currently adopt OpenClaw due to security and compliance issues, meaning the market for serious enterprise clients is years away. He contrasts this with "building on a rock," advocating for a proven business model: an AI auditing and consulting agency.

Ottley argues that the agency model addresses a fundamental "human problem" in AI adoption—the lack of skilled people to identify business needs, build and deliver solutions, and manage team training and change management. This is a slow-moving but stable problem that will persist for the next decade, offering a much longer horizon for value creation. The skills gained in auditing, solution delivery, and training are universally valuable and transferable, regardless of new platforms emerging. He concludes that for most people, especially beginners, the "boring, proven, and slower moving opportunity" of an AI agency is a far smarter and more stable bet than chasing the flashy but unproven "skills" market, offering a free course for those interested in this more reliable path.