AI is already deeply integrated into our everyday communication—it’s trained to detect grammar mistakes in Google Docs, suggest predictive text in iMessage, and handle customer service requests. Thanks to major recent advances in the sector, everyday use cases are on the verge of a boom, drawing a lot of attention within the space right now.
In fact, the AI-powered writing software market is expected to jump from $372 million in 2021 to nearly $1 billion by 2030.
While there’s certainly a lot of hype circulating the sector at the moment, we’ve been tracking generative AI’s potential in communications. Much of my work at Headline involves tracking startups’ website traffic using our proprietary internal software, Searchlight, to identify new areas for investment. In 2019, we began to notice that AI-generated communications tools were gaining huge adoption rates, more than any other prosumer SaaS products, building up the early stages of our conviction toward the sector.
This trend is only growing as we see new areas of AI in communication, from helping you spark up a conversation in a dating app to creating original images.
It’s clear users are ready for these tools now that they are working well. Their potential was largely sparked by the launch of GPT-3, an evolution in software released by OpenAI in 2019 (the same time we saw an uptick in traffic). GPT-3, an autoregressive language prediction model, added a level of sophistication to AI tools by being able to generate human-readable text. When leveraged correctly, it not only empowers people to write better but also allows them to step back and become the composer of a piece while AI generates text and a general flow to work with.
On the surface, using AI to correct grammar in a research paper and leveraging it to draft a dating profile sound quite different, but they come from the same line of evolution. And we’re just tapping the surface of what’s possible.
The Past: Using AI to Cross T’s and Dot I’s
Founded in 2009, Grammarly is the undisputed behemoth in the AI-assisted writing tool space, and where it all began. Its original concept was simple: improve people’s writing by correcting their grammatical, spelling, and stylistic mistakes. By 2015, the time Grammarly released a free browser extension so that anyone—students, writers, professionals, or even just friends emailing back and forth—could access it, the tool had one million active daily users.
In just five years, that number grew to 30 million as Grammarly’s abilities expanded to include identifying a writer’s tone, suggesting words to use for stronger sentences, and ways to form more coherent paragraphs.
Figure 1: Representing Grammarly’s 10 year website traffic within Headline’s Searchlight platform. The X axis represents years up to 10. The Y axis represents site ranking up to 90k visitors. (The Searchlight Growth Index is a composite of third-party data sources around a company’s website traffic. The Y-axis is based on rank, with Google being ranked number one as it is the domain with the most traffic on the observable internet. Each color segment represents a magnitude of traffic, ranging from dark blue—highest ranks, little traffic—to dark red—lowest ranks, most traffic. The vertical green lines in the charts represent a point in time where the company raised outside capital.)
At the same time, other tools began to crop up as competitors that are still going strong. Founded in 2013, Hemingway App makes someone’s writing more clear not only by copy editing their passages, but also by suggesting more concise ways of expressing convoluted ideas. That same year, ProWritingAid was born in Europe offering a similar solution.
Figure 2: Representing Hemingway App’s 10 year website traffic within Headline’s Searchlight platform. The X axis represents years up to 10. The Y axis represents site ranking up to 160k visitors.
Figure 3: Representing ProWritingAid’s 10 year website traffic within Headline’s Searchlight platform. The X axis represents years up to 10. The Y axis represents site ranking up to 130k visitors.
The Present: Turning Writers into Composers
Once more tools started to crowd the sector, the release of GPT-3 gave companies the ability to differentiate their offerings, which is when solutions evolved to support users not just write and edit content, but also compose it by pulling from work the AI generates and structuring it in their favor.
Figure 4: Representing QuillBot’s 10 year website traffic within Headline’s Searchlight platform. The X axis represents years up to 10. The Y axis represents site ranking up to 3M visitors.
Next on the horizon is actual content creation—meaning that writing from scratch is quickly becoming simpler and perhaps in the future, even unnecessary at both the professional and personal level. Keys AI, for example, is a simple iOS app that reads screenshots of messages and suggests responses. Think about the stress of coming up with conversations on dating app. Keys AI suggests text with the intent of carrying on a conversation. Products like Regie.ai and Lavender.ai bring sales teams solutions for crafting outreach emails to prospects, generating copy backed by data.
Figure 5: Representing Keys AI’s 10 year website traffic within Headline’s Searchlight platform. The X axis represents years up to 10. The Y axis represents site ranking up to 3.2M visitors.
Figure 6: Representing Lavendar.ai’s 10 year website traffic within Headline’s Searchlight platform. The X axis represents years up to 10. The Y axis represents site ranking up to 38k visitors.
The Future: Expanding AI into Audio and Beyond
AI has gone from simple writing suggestions to creating content on its own, and now, it’s moving outside the bounds of text. Headline portfolio company, Avoma, founded in 2017, is a perfect example. In real-time, Avoma transcribes conversations into text with incredible accuracy. At Headline, Avoma has become integral to the way we work as we hold meetings with founders. Where Avoma really shines: not only transcribing text but understanding the context of the conversation in order to provide actionable coaching insights. We share these learnings with the rest of our team, which enables us to get everyone up to speed faster and build conviction about deal opportunities.
By bridging the gap between written and voice, Avoma is ultimately creating a new frontier of products that can leverage AI to not just augment how we communicate through text, but also how we speak.
Figure 7: Representing Avoma’s 10 year website traffic within Headline’s Searchlight platform. The X axis represents years up to 10. The Y axis represents site ranking up to 40M visitors.
AI is also moving into visuals. DALL·E 2 is OpenAI’s latest AI system creating new waves of capability in this space. Users simply type in a description—say, two robots eating dinner on the moon—and the platform generates a realistic image from the natural language.
Figure 8: Representing OpenAI’s 10 year website traffic within Headline’s Searchlight platform. The X axis represents years up to 10. The Y axis represents site ranking up to 350k visitors.
Thanks to the ever-increasing quality of AI-assisted communication tools, people will find themselves needing to write from scratch less over the next decade. For all content, we’ll soon become the director overseeing the process—rather than the actor on the ground. What’s still to be determined: what emerging technology will be embedded features in existing products versus what holds enough weight to stand on its own.