Low Hanging Fruit but Slim Pickings: AI for Copywriting

Where's the moat? When your platform can be wiped out by big company adding a feature, you don't have a sustainable business.

The Backstory and Landscape

The landscape of AI-assisted copywriting and marketing tools has quickly grown into a dizzying array of platforms, and if you try to choose, you’ll find little differentiation. Maybe one integrates with a tool you already have, maybe one has SEO optimization, etc. but not much else to look at.

When I set up my venture notarybrothers.com, I looked to AI (specifically Jasper.ai) to help create the sales and marketing copy - emails, website text, video scripts, blog posts, etc. I quickly found it more trouble than it was worth. The copy wasn’t of the highest quality and I spent a great deal of time editing. That was until ChatGPT came out. ChatGPT is so flexible and easy to use that I quickly found myself using it for everything.

It's clear that automating the tedious work of writing copy and marketing material is a natural first application for AI. With each generation of OpenAI’s releases (GPT-2, GPT-3, and now ChatGPT/GPT-3.5), there’s been a huge leap in writing quality, and now the entire world knows - copywriting is going to be almost totally automated, along with many other forms of writing.

I think everyone woke up to the potential of AI-assisted writing when GPT-2 came out in 2019. I remember being amazed at a simulated subreddit controlled by GPT-2 and how often it would create high-quality content.

In the few years since GPT-2 was released, a plethora of startups in the AI copywriting space have emerged, such as Jasper.ai, Copy.ai, Rytr, Writesonic, Wordtune, Anyword, Lyne, CopySmith.ai, the list goes on and on. I will say that after examining and testing many, there is little differentiation between them. I think most founders of AI-assisted copywriting and marketing tools did not expect the rapid improvement and proficiency of AI engines, which has made the efforts of these startups in complicated prompt engineering for older AI engines obsolete.

What differentiators do copywriting AIs boast?

1. AI engines. The vast majority use GPT, but some, like ClosersCopy, use their own engine.

2. Templates and workflows. Almost all the startups boast the number of templates they offer, such ones for creating blogs or Amazon product descriptions. There are literally hundreds of other use cases.

3. Integrations, APIs, and plug-ins with existing platforms. Think Grammerly’s plugin for Chrome, or Zapier integration, or Jasper’s partnership with Surfer SEO.

4. Built-in SEO tools and analytics. Some include tools for optimizing writing for search engines and analytics capabilities.

5. Post-processing tools. Plagiarism checking, sourcing with links, etc.

6. Bulk creation. For example, writing multiple blog posts at once, or translating simultaneously into many languages.

However, none of these differentiators build a sustainable competitive advantage.

Why do these differentiators not matter?

1. AI engines are plug and play. Right now ChatGPT is the best in the room, and soon more are coming out. There is certainly an advantage to having the best AI engine on your platform, but every sane competitor will move over as well. A startup using their own AI engine is delusional if they think they’re going to be better than the ones created by OpenAI, Google, Meta, etc, which have 100’s of billions of parameters and are trained on damn near the entirety of the internet.

2. Templates are essentially just forms slightly customized to the relevant workflow along with prompt engineering (“Take user input and write a cute blog post title in <60 characters that is compelling to readers and is good for SEO”). They’re a dime a dozen, and usually the workflows they are automating are extremely simple. I personally found using ChatGPT to get what I want and tune easier than using predefined forms. Companies like Copy.ai boast over 100 templates. One engineer can set up multiple templates in a day.

3. Integrations, APIs, and Plug-ins aren’t a sustainable competitive advantage. There may be some small winners here, especially if they integrate well (looking at you Grammerly), but what makes more sense is a product that already has a moat (e.g. Adobe Photoshop) adds their own AI feature. Being a plug-in is a tough business model, especially if you’re building something really difficult, like a new Plaid.

4. Built-in SEO tools and analytics are easy to build and replicate, and there are already a million SEO startups out there that can just as easily add AI content writing.

5. Post-processing tools like plagiarism and sourcing are much-needed, but Microsoft/OpenAI and the other big players are well-aware. They’re building them into to the next iterations of ChatGPT and others, meaning this differentiator will be gone before the end of 2023.

6. Bulk creation is very easy to implement. Just generate more queries for the AI engine.

So who will own the copywriting game?

Easy - the big guys. It's clear that the major players in the tech industry will ultimately own the copywriting game. These companies already have significant resources and established moats, and they are quickly adding AI writing and creation features to their brand-name products. This creates a smooth user experience with little friction, as there isn’t a need to switch between different tools.

These large companies have been slower to fully embrace AI-assisted writing due to ethical considerations such as plagiarism, sourcing, and “problematic” content. However, they are currently working to develop reliable processing and post-processing tools and are still releasing AI writing features at a rapid pace. It's likely that they will continue to dominate this space in the future.

Some examples:

• Gmail has had auto-complete for emails for years as feature called Smart Compose. There’s no way they’re not going to release the tool in a more comprehensive form.

• Mailchimp already has AI creation on their platform, and can very easily add more comprehensive AI-assisted content writing. Again, this is something I’m sure they’re working hard on.

• Microsoft has already been deploying AI to Powerpoint, and is definitely looking at doing the same with Word, Excel, OneNote, Outlook, their entire suite of products, especially since they partially own OpenAI.

• Adobe has already been integrating generative AI into Photoshop and it’s only a matter of time before content writing is integrated into their other products like Illustrator or InDesign.

The Takeaway

Generative AI tools are proving to be features, not products. It’s extremely difficult to create a sustainable advantage when your product can be taken out by a platform adding a feature.

In the end, it is likely that the big players with already established moats, such as Adobe or Microsoft, will dominate this space as they have the resources to easily incorporate AI writing and creation features into their brand name products. As such, angel and VC investors should be wary of investing in startups in this space, as the barriers to entry are low and the competition is fierce.