The Birth of AI-powered FastSaaS

GenerativeAI is the rage right now, and companies are waking up to this less effort, more returns bid for the future. The current wave of SaaS products is developed around AI use cases ever since the release of LLMs like OpenAI’s GPT-3 and diffusion models as commercially-available APIs. 

In recent times alone, Shutterstock said it would integrate Dall-E 2 to its platform, Microsoft VS Code added Copilot, the AI pair-programming assist, as an extension, TikTok announced the rollout of an in-app text-to-image AI generator, and Canva came out with an AI-generating feature on its platform. 

Not to forget, a few days ago, Notion released its own AI domain which integrates AI-generated solutions to its all-in-one workspace. The new launch revealed that, with NotionAI, users would be able to generate “any type of content”—be it writing blog posts, brainstorming new ideas, creating press releases, ideating creative social media posts, summarising and translating texts, among several other things. 

It is important to note that these large-scale enterprises are up against plenty of startups that have raised funding to build AI-powered tools. Notion, for instance, is competing with Jasper, which raised $125 million for its AI content platform, and Instoried, the content creation startup which raised $200 million in funding, and Regie.ai, which raised $10 million for its business content creation platform. The list is endless. 

This list seems incomplete without Mem. Led by investment from OpenAI Startup Fund, Mem was the first to introduce an AI-powered workspace organisation tool, with its product, Mem X. Notion’s reinvention of its platform may have been a direct effect of the shift in tide towards many of the AI application tools.  

Same same, but different

All the startups mentioned above use GPT-3 for content generation. It is not clear yet if NotionAI also uses GPT-3, although speculations suggest it does. The question here is: If all content/image tools are based on GPT-3 or Stable Diffusion, what really separates one from the other? 

Chris Frantz, a co-founder of Loops, addresses this, saying, “There’s almost no moat in generative AI. That means that the differentiator is product, marketing, use case and time to market.” That is, with every SaaS company boarding the AI train, the only differentiating factors between them is the interface, the ease-of-use, accessibility, and the ability to leverage the existing user base on the platform. 

So, even if large enterprise companies are not the first-movers in terms of leveraging the AI fever, they have the advantage of having a substantial user base to begin with. NotionAI, for example, had a waitlist for over 50,000 users, and this was only a few hours after the launch. 

FastSaaS: New model of product development

Therefore, it is safe to say that we are entering into an age of, what Matt Krandel calls, ‘fastSaaS’. The term represents how AI and no code are lowering barriers for developing software products. This, Krandell says, “can lead to margin compression, and more importance on distribution”, similar to what we mean when we say ‘fast fashion’.

An interesting aspect to notice also is how a lot of the companies using AI are diversifying their product portfolio—not to bring more cutting-edge tech onto their platform, but to just have more and more AI-assisted features integrated into them. If we simply compare the AI copywriting tools available in the market today, we notice an abundance-in-making. Each tool segments into multiple functionalities. 

Writersonic, for example, labels multiple use cases generated to the same writing API like articles and blog posts, ecommerce product descriptions, ads and marketing tools, website copy, and numerous other general applications. AI copywriting tool Simplified also has an AI-image generator to the list of features it provides with an AI writing tool. 

And with Notion, we are seeing services not only related to AI writing, but also additional ones that include spellcheck and grammar, paraphrasing, or translation tools. Foraying in SaaS at the moment is like entering an arena of excess production, with little or no understanding of consumer needs. 

https://twitter.com/mattyhogan_/status/1593272623378710529

Beyond LLMs as AI generators

Basically, building a SaaS is made equivalent to creating an “everything” app. A Twitter user by the name ‘Linus’ makes an interesting analysis of the case. Linus says that companies should be thoughtful about integrating AI into their product. Large Language Models (LLMs) are not merely generating machines, but have human-like reasoning capabilities, and products should thereby be modelled around it—something which was also iterated by Alexandr Wang, CEO of Scale AI. 

Linus writes that the real deal is not in generating texts, but in “using this new capability and wrapping into jobs that have other shapes”, while also adding that, “text generation is not the product, reasoning is”. 

Speaking about NotionAI, Linus stressed that the product, if remains limited to text generation, will end up making “10 copies of every piece of information, 4 of which contradict each other and only 2 of which reliably surface on searches”. Hence, beyond its application to narrow workflows, the content generating machine can leave organisations with more information than necessary.