Beyond the AI Hype: Are Startups Just Repackaging APIs? Reflections from TechCrunch Disrupt 2024
Saad Benryane
At the close of Day 1 at TechCrunch Disrupt 2024, one thing is abundantly clear: AI is dominating the conversation. From customer service automations to industry-specific solutions, nearly every startup has a pitch that revolves around “AI-driven” capabilities. But here’s a twist—many of these “AI products” are actually just interfaces overlaid on pre-existing AI models, primarily APIs from companies like OpenAI , Google’s Bard, and Anthropic’s Claude. Startups leverage these APIs to deliver fast results, but without proprietary models, the innovation risks feeling superficial, more branding than breakthrough.
By relying on these external APIs, startups get to market fast, sidestepping the enormous costs of developing their own models. However, this dependency means that if OpenAI, Google, or Anthropic adjust their pricing or launch similar features, these API-dependent startups face significant challenges. The fragility of relying on another company’s tech for core functionality has already proven risky, with some “wrapper” companies seeing their value proposition suddenly compete with their foundational provider’s own new features.
The Appeal and Challenges of Using Pre-existing AI APIs
There’s no denying the appeal of APIs like OpenAI’s GPT or Google’s Bard. These tools are robust, allowing companies to incorporate advanced language capabilities quickly. Startups like Jasper and Copy.ai initially gained traction by wrapping OpenAI’s models with user-friendly interfaces and customizable templates for content generation, serving niche markets without massive in-house AI development costs. The rapid time-to-market made it a winning strategy, but as competition floods in, the challenge of differentiation grows. Furthermore, costs remain closely tied to API pricing, limiting how much these startups can control their own margins.
For startups trying to scale, data privacy and security add another layer of complexity. Industries like healthcare and finance, where data is highly sensitive, demand compliance with strict data sovereignty laws. For instance, Morgan Stanley uses OpenAI to empower wealth managers with rapid information searches across large content repositories, but all within stringent internal controls. Stripe, a financial tech company, uses AI models to streamline customer support and monitor fraud, integrating GPT-4 to enhance these workflows while addressing security requirements. These examples show how businesses can innovate while prioritizing user data protection.
AI as Quiet Infrastructure, Not the Headline
The AI startups that truly stood out today are those that use AI quietly, enhancing core functionality without overselling it as a primary feature. Like the essential but unseen components of a tech stack—databases, cloud storage, or cybersecurity layers—AI is most effective when it operates as an infrastructure element, powering meaningful, differentiated features without overshadowing them.
For instance, Morgan Stanley’s AI model isn’t branded as a standalone product; it functions as a resource within a larger ecosystem, empowering advisors with real-time insights without making AI the centerpiece. Similarly, Stripe integrates AI for fraud detection and customer service improvements, allowing these capabilities to run seamlessly in the background. These approaches highlight how AI can add real value when it powers applications that address specific user needs without claiming center stage.
A Growing Ecosystem of AI Providers and Competitive Choices
Fortunately for founders, the landscape of available AI models is expanding. In addition to OpenAI, companies like Google, Anthropic, and Cohere are developing competitive large language models, giving startups more options to choose from. This diversity of offerings helps mitigate dependency risks, keeping costs manageable while allowing startups to experiment with different APIs to suit specific use cases. As competition grows, pricing will likely stabilize, providing startups with more leverage to choose the best technology for their goals.
This competitive environment also enables startups to build unique AI solutions tailored to industry-specific challenges, creating more space for innovation that goes beyond simply wrapping an API. By selecting the most suitable API for their needs—or even using multiple providers in tandem—companies can prioritize functionality, reliability, and cost-effectiveness.
The Path Forward: Focusing on Real, Impactful AI Solutions
AI’s real value lies in its ability to solve specific, pressing problems. While API wrappers can offer quick solutions, the future belongs to startups that use AI to solve meaningful issues. This approach involves understanding industry-specific needs and designing AI tools that improve productivity, inform decision-making, or automate complex processes.
Take Openfair (shameless plug), for example, which supports entrepreneurs, family offices, and private equity firms in analyzing thousands of business documents quickly. This level of rapid analysis helps clients make more informed decisions in business acquisitions, saving time and enhancing accuracy. However, ensuring data security and compliance is paramount, so Openfair is prioritizing data sovereignty before scaling its AI-powered solutions fully. It’s an example of AI designed to solve practical problems while safeguarding user data.
This era of experimentation is essential for AI’s evolution, encouraging companies to push boundaries and create tools that add real value across industries. While some startups will continue leveraging pre-existing APIs, those that seek to address deeper needs will likely set the standard for future AI-driven innovation.
Disclaimer: In a twist of “AI-ception,” this article—reflecting on AI overuse—was crafted with a bit of help from AI itself.
