GrammarlyGo and the coming wave of generative AI productivity

Generative artificial intelligence (AI), particularly ChatGPT, has taken an AI market that seemed far off into the future and quickly catapulted it into our world. But the tools still seem raw, more suited to coders than end users who want a more worker-friendly productivity tool.

That changed this week with the announcement of GrammarlyGo. Since I’m a Grammarly user, I’m particularly excited about this addition and I expect it to give us a frame of reference for when Microsoft eventually adds generative AI to Office. 

Generative AI could do much for those of us who write a lot and deal with a never-ending stream of email. 

The GrammarlyGo move

GrammarlyGo is a prompt-based implementation of ChatGPT (though it could use other frameworks in the future). If you are writing a paper or a column like this one, it will ask you a series of questions —  and from the resulting answers, it will write the piece faster than you could type it. The questions it asks are about tone and content and what specific data you want included, and the result is something that appears to have been custom-written by you. 

It won’t capture your personal style well until it learns how you write, or until it writes enough content that it becomes the way you write. I expect future versions will be better able to tell context and, from your historical writing, be able to better emulate exactly how you would write a piece. 

Grammarly took me through a couple of demos; this is what I observed.

Starting an email

If you’re originating an email thread, it will ask you what you want done and for any specific information you need included before generating the email. You can then make selections that change the email’s tone as needed and ensure what it writes is what you intended. 

I immediately thought of my old friend Steve Ballmer, Microsoft’s former CEO, who used to write emails as if each word caused him pain. A two-page email to him would get often get a one- or two-sentence reply — if you he responded at all. This tool would have allowed Ballmer to do his limited-response thing, but still deliver a far more complete email. Generative AI could look at an email sent from someone else and craft a recommended response, allowing Ballmer (or anyone who hates writing emails) to appear much more empathetic and engaged without significantly increasing their workload.

(I have no doubt it would have improved how people that corresponded with him.)

In addition, the tool can craft survey questions, bullet lists, or other richer forms of email on request. 

Ending writer’s block

Most writers have days where we look at a blank document and get stuck. If you can at least describe what you want to do, GrammarlyGo will make an attempt to get you started and remove that writer’s block. It can even generate lists of topics to consider if you aren’t sure which topic to cover. I personally find condolence letters hard to write, and the thank-you notes I should send usually end up unwritten.

Grammarly Go

GrammarlyGo can do those automatically. In short, for much of the regular writing we do, it could serve as a helpful tool, handling writing tasks we’d prefer to avoid or, at least, creating a draft we can use as the foundation for our own work. 

Depending on how bad your writer’s block is, this could shave days or even weeks off a particular project and allow you to meet deadlines. 

AI is just getting started

This is still just the leading edge of the generative AI journey. It will no doubt move on from audio to eventually embrace video production. (The level of potential disruption in a variety of issues is through the roof.) Workers who think they might soon use this kind tool should train for it aggressively, which is key to preventing the need for massive layoffs once the technology matures. It’s also important to note that with generative AI comes a host of legal and ethical issues about content creation and ownership that are nowhere near being solved.

Before long, most everything we interact with will have some kind of generative AI behind it. This could result in a better work/life balance because AI tools can take over some of the most annoying and repetitive parts of our jobs, help us think through problems, and turn out creative works that excel in both quality and quantity.

GrammarlyGo may be one of the first solutions of this kind, but it will be far from the last to make this jump.