Summary of Key Points
In the age of AI, software entrepreneurship may seem like an opportunity with lower technical barriers (large models can write code quickly and at low cost), but the actual environment has become more challenging: financing is difficult, markets are shrinking, and competition is fierce. Many entrepreneurs face difficulties such as price wars, slow revenue collection, and products being replaced by large models. However, opportunities still exist—real opportunities lie in the "industry expertise" and "specific business processes" that large models cannot handle. Entrepreneurs need to focus on a single industry, achieve tangible business results first, and then standardize these experiences into products to survive and thrive in the current environment.
Detailed Analysis
Why is software entrepreneurship so difficult in the AI era? Real-world examples provide the answers
Many AI entrepreneurs are facing tough times:
- Price wars erode profits: Some companies are seeing growth in both products and customers, but price wars undermine their profit margins, and financing delays leave them with dwindling cash flows.
- Custom projects drain team resources: Companies working on intelligent assistant projects for state-owned enterprises generate revenue but have slow payment collections. They spend all their time on customizations and never have the time to develop standard products, leaving them struggling without a clear future.
- Small teams struggle to grow: Small teams with limited funding can only move slowly; they neither fail immediately nor grow rapidly.
- Product obsolescence due to model updates: Teams that develop intelligent assistants see their efforts go to waste when large models release free versions that offer similar functionality.
The reasons for these challenges are three major changes in the environment:
- Stricter financing requirements: Capital used to invest based on revenue and customer base; now, investors also demand profitability and cash flow, questioning whether a company will not be replaced by large models or established firms.
- Shrinking market demand: Ten years ago, small and medium-sized enterprises were willing to pay for new tools, but now business owners are cutting back on IT investments, focusing only on short-term returns, leading to the loss of key customer groups.
- Fiercer competition: Large consumer-facing companies (e.g., AI chatbots) and top SaaS providers (e.g., customer service, sales systems) have significant market presence and channels, making it easy for startups to be surpassed. With continuous model upgrades, the lead time for generic products is shrinking, and customers may see your product as a temporary solution.
How much easier was software entrepreneurship ten years ago? Compare it to the SaaS era
Ten years ago, when SaaS (Software as a Service) was just emerging, the entrepreneurial landscape was very different:
- Customer-friendly: Small businesses had no historical baggage and were open to trying new technologies. Early SaaS solutions were simple but offered good mobile experiences at low prices, quickly gaining market traction.
- Capital support: Investors were patient; companies with annual revenues in the tens of millions could secure hundreds of millions in funding. Examples like Youzan and Beisen emerged during that period and are now generating stable annual incomes in the hundreds of millions.
Back then, the market provided space, and capital gave time for entrepreneurs to refine their products. These factors no longer exist today.
Where do the opportunities lie? In industry expertise and business processes that large models cannot handle
Enterprise AI can be divided into three layers:
- General intelligence layer: Large models (e.g., GPT, Wenxin Yiyan) can perform 80% of general tasks well but struggle with more complex ones.
- Industry-specific knowledge layer: Rules for diagnosis in healthcare, lesson preparation methods in education, and manufacturing standards—these are industry-specific practices that large models do not understand.
- Enterprise-specific processes layer: How a company organizes its work, manages price wars, and deals with difficult customers—are unique to each business and cannot be learned by large models.
Customers are truly willing to pay for solutions that solve practical problems (e.g., increasing sales or reducing the need for manual customer service), not just for cool-looking demos. Therefore, entrepreneurs' opportunities lie in leveraging industry expertise and business processes that large models cannot overcome.
Survival guide: Don't rush; focus on one thing at a time
The most realistic path for current AI entrepreneurs is to:
- Step 1: Focus on an industry and achieve tangible results: Choose an industry and address specific customer problems (e.g., helping sales teams close more deals). Don’t just focus on launching the system; show customers concrete benefits.
- Step 2: Develop a delivery methodology: After completing a project, summarize your experience—how many stages are involved? What needs to be done at each stage? What are the risks? Turn these into standard processes to ensure success in every project.
- Step 3: Standardize products: Repurpose functions that appear repeatedly in projects into standard products. This reduces costs for future deliveries. Companies without standardized products will always be limited to custom work and have no long-term prospects.
Remember: Crossing industries to take on projects may seem like a way to increase revenue, but it prevents the accumulation of valuable experience. Expanding the team too early can further strain cash flows. The key is to master one industry and one project thoroughly before moving on.
In conclusion
In the AI era, software entrepreneurship is not without opportunities; it requires a shift from competing on technical speed to focusing on industry depth. By capturing the details that large models cannot handle and doing things well, entrepreneurs can survive and thrive.