Grammarly acquires Superhuman
110 comments
·July 1, 2025lvl155
Grammarly has an existential crisis. It can be replaced it with free versions of the top models and they are much better (and I can control the UI anyway I want). In fact, many of these “web 2.0” business models are a few more updates away from getting replaced.
myflash13
This is such a programmers take. Bottled water is in a crisis, it can be easily replaced with tap water and a reusable container! Yet it’s a $47 billion dollar market in the US.
Grammarly’s value is not in having a replaceable product, it’s in the network, distribution, customer acquisition channels and integrations with tools. Like bottled water, it’s about being in the customers face at the right place and the right time.
lvl155
You’re on the money and that’s exactly why I think Grammarly will struggle. OpenAI/Gemini/Claude will get embedded further. Gemini is already on gmail. Getting OpenAI and Claude incorporated is trivial. Guess what? Once Apple figures out what they’re doing with AI (which I hope is to buy Anthropic) they will take whatever is out there and incorporate them into iOS/MacOS just as they’ve done for so many third-party app ideas in the past.
StochasticLi
I'm paying for most AI models top tiers and Grammarly. Grammarly is a phenomenal tool. It's not that LLMs can't do it. Well, they can't, but the more important thing is Grammarly's UI.
JoeDohn
not to mention languagetool
ignoramous
Grammarly can cut down their costs if they use those models themselves. The current LLM advancements aren't disruptive but incremental. What's the hurdle you see they can't rely on their existing distribution and expand from there?
roguecoder
Why would using a more-expensive technology cut their costs?
rwmj
It's the dot-com strategy. They'll lose money on every sale but make it up in volume.
lvl155
That’s true but I can replace Grammarly in a few hours with Claude. 99% of the functionality. Then, I can spend a couple more days to add stuff that they can’t add due to copyright.
StochasticLi
Grammarly has a function that replaces phrases that are typical for LLM output. Good luck coding that in a few hours.
micromacrofoot
yet they recently received a billion dollars in financing this year?!
jgalt212
Yes, does not seem like such a crisis to me.
micromacrofoot
maybe not that crisis, but perhaps a crisis
briandoll
I've been looking to replace Superhuman recently. None of their AI or Team features matter to me. I just wanted what they originally set out to build -- a super fast, keyboard driven, desktop email client. There are daily paper cut bugs and search issues that have persisted for many years, and I'm not going to stick around through this transition which will surely make the product worse.
What do folks like for desktop email that's keyboard driven? At this point I almost want to go back to Pine ;)
myflash13
Same here, I recently got frustrated with every email client, tried everything. Few days ago I finally decided to vibe code my own email client with Claude Code and I got a basic version running in a single day. Can’t wait to build it exactly the way I want, with programmable rules/filters and AI drafts for specific types of emails I get, conversion to plain text or dark mode for readability, contextual information in sidebars pulled from APIs (such as email history and customer support / CRM info), and one click actions in other apps.
rwc
Mimestream on Mac implements GMail's keyboard shortcuts so there was no learning curve for me and I'm able to enjoy a desktop app experience.
briandoll
This does seem pretty seamless to swap to, thanks!
Now to figure out iOS ;)
swyx
chiming in with thanks, i was looking to leave Superhuman also.
shepherdjerred
Mailmate on macOS is good
umbra07
I use (neo)mutt
dgellow
Oh no. I love superhuman :( Please don’t change it
JSTucker
> The company claims its users send and respond to 72% more emails per hour, and the percentage of emails composed with its AI tools has increased fivefold in the past year.
Is this really a good metric to aim for? Don't we want productivity tooling to result in less email not more?
Aurornis
Anecdotally: The Superhuman users I've worked with start skimming e-mails and sending super-short replies. Sending a "Good job team" or questioning something in sentence 2 that would have been answered if they read all the way to sentence 5 of the e-mail is the way to clear their inbox.
The inbox->outbox flow turns into the way to clear the inbox. It's not about better communication, it's about speedrunning their way to inbox zero.
The worst case was a person who would respond to everything with a one-sentence question, then respond to the response with another one-sentence question, and repeat all day long. He could turn a brief e-mail into a thread with 15 one-line responses that could have been avoided by spending more than 10 seconds thinking about it.
cryzinger
This old-ish Newport essay comes to mind:
> The knowledge sector’s insistence that productivity is a personal issue seems to have created a so-called “tragedy of the commons” scenario, in which individuals making reasonable decisions for themselves insure a negative group outcome. An office worker’s life is dramatically easier, in the moment, if she can send messages that demand immediate responses from her colleagues, or disseminate requests and tasks to others in an ad-hoc manner. But the cumulative effect of such constant, unstructured communication is cognitively harmful: on the receiving end, the deluge of information and demands makes work unmanageable. There’s little that any one individual can do to fix the problem. A worker might send fewer e-mail requests to others, and become more structured about her work, but she’ll still receive requests from everyone else; meanwhile, if she decides to decrease the amount of time that she spends engaging with this harried digital din, she slows down other people’s work, creating frustration.
https://www.newyorker.com/tech/annals-of-technology/the-rise...
I'm hesitant to call the email-skimming workflow that you mentioned a "reasonable decision," but I think the point still stands about how one person speedrunning their inbox can make everyone else's inboxes that much worse.
Vegenoid
“Increased volume of email” sounds like something people would pay to avoid.
rchaud
That's how companies like Slack get billion-dollar valuations. The promise of "less".
pchristensen
There are many people whose job revolves around churning through emails (sales leads, recruiters, etc). This is a huge win for them.
chii
> percentage of emails composed with its AI tools has increased fivefold in the past year.
read: spam has increased 5 fold!
mattcantstop
I think Superhuman's CEO in an interview said their product is specifically catered to people who are seeking inbox zero.
For those people this would be a great outcome. The question is should this be the goal of most people? Probably not. But most people are not their ideal customer. They explained their ideal customer in depth in an episode of the Acquired Podcast.
mrweasel
A better metric would be: How frequently does the recipients of those emails need to reach out for clarification. The goal of any writing should be to increase clarity and ensure that your message is clearly received.
Why do their customers even need to send 72% more email?
cik
I find this particularly fascinating, given the post email world I live in now. I haven't had an email from a contact in over two years. It sounds like a sales tool, in a world where the goal is to distance from that availability.
saaaaaam
Was this “post email world” a choice or something that happened? How do you communicate rather than by email now?
cik
In my country, e-mail is now used purely as a sales tool. Notifications from government and schools are also delivered via SMS, and either WhatsApp or Telegram. So.. yeah, block everyone.
Turns out the second you do this you eliminate 100% of the spam in your life. Honestly, if I ever lived in North America again, I think I'd also just stop reading e-mail.
apparent
Yeah, if they could increase the quantity without affecting the quality (or improving it), that would be great. But there's a good chance that is not what's happening.
chaosprint
This is a bit surprising. I even didn't expect Grammarly to have the cash, I used to be a paying customer of theirs when I was writing papers, but apparently with AI I don't even need the free Grammarly anymore.
icey
They might have a lot more cash than you'd expect: https://www.grammarly.com/blog/company/grammarly-announces-g...
swyx
and superhuman might be worth less than they raised at - https://x.com/pitdesi/status/1940079704423506401 (or not, we simply do not know)
solumos
The undisclosed acquisition price/terms implies a down-round exit at best, and a fire sale at worst.
rcleveng
That was my thought as well, but one of my college age kids still likes Grammarly over just using ChatGPT for grammar checking and rewriting, says it does a better job.
Excited to see what they are doing now after the "acquisition" of Coda (seems like a bit of a reverse acquisition or acquihire since they buy Coda and have Coda leadership take over Gramarly.
FireBeyond
I would actually agree with that, too. Grammarly certainly isn't perfect (it still occasionally struggles with the nuance of some idioms or proper nouns), but it does better than LLMs (I use MacWhisper with its "CleanUp" AI prompt for dictation). But Grammarly's inline use is actually pretty handy (even in this text box, I pause typing for a moment, and there's a Tab prompt that will auto-edit my text live).
anilshanbhag
Grammarly is one of the tools I pay for, and I am worried about the security risks of using it. Really wish there was an alternative that: 1) Does local processing (local LLM?) instead of sending all my data to their server. 2) Had a lightweight Chrome extension that didn't inject many MBs of scripts on each page.
nwjsmith
Harper checks a lot of your boxes and is getting better all the time: https://writewithharper.com/
jgalt212
How does Harper compare with LanguageTool. We use a privately hosted version. It's better than nothing, but in practice it's more like a super-charged spell checker.
hiatus
On the homepage it has a comparison of figures, presumably indicating response time, though it doesn't speak to its performance in terms of grammatical errors caught:
Harper - 10 ms LanguageTool - 650 ms Grammarly - 4000 ms
jakub_g
FWIW: latest Chrome ships built-in AI APIs
https://developer.chrome.com/docs/ai/prompt-api
so it should be a matter of time to have a replacement extension using this local API. However the built-in model is Gemini Nano.
swyx
yeah i highly doubt people will use Nano for more than simple retitling because we're so used to higher intelligence for ~free elsewhere
treetalker
For a more-classic, more-human experience (i.e., computer flags potential issues, you decide and correct if necessary) there are proselint and vale.sh.
diggan
Feels like that'd be trivial to build, biggest issue is having to ship large files (LLM weights), but maybe CNNs would be enough, I'm guessing Grammarly started with CNNs or similar?
What are you using Grammarly for, is it just spell/grammar checking or something more? Is the UX particularly good? Personally I tried it some years ago but didn't understand/see what is/was special about it.
NewsaHackO
I personally used to have a subscription for grammar checking, especially for longer papers. Now, I just use a LLM. I personally don’t see the strategic value of them pivoting to using genAI; there is no way I would pay $30 a month for something that will take at most 100k tokens using other LLMs. They seemed to have heavily downplayed their unique aspect which is the deterministic ruleset.
voigt
> Superhuman valued at $825 million in 2021, $35 million annual revenue
This is nuts! I used Superhuman for about a year. And honestly, I might still be using it if the pricing weren't so off. It had a couple of nice features, and the keyboard-driven approach was a welcome change for mail clients.
But ultimately, Superhuman had nothing that couldn't be replicated in a relatively short amount of time (maybe even with plugins?).
$825 million? Maybe I should start a mail client company...
saaaaaam
Zoom was worth around $125 billion at the start of July 2021. It’s now worth about $23 billion.
So by that logic, Superhuman may be worth around $165 million.
More interestingly though, let’s assume they spent the $110 million they raised. That means that each of the ~85k customers they would appear to have based on the estimated revenue cost them about $1300. Though probably more as a proportion of ongoing revenue will obviously be driving sales and retention.
I did see something somewhere saying that they have very high customer retention. That matches my anecdotal experience - I’ve been using it for several years as have several people I referred.
But yeah… an $800m+ valuation? That feels like Covid-era hype.
harry2quinn
This feels like a pattern of grammarly becoming a holdco / following the Salesforce playbook. Find companies with - solid but not breakout growth (and probably slowing) - a loyal cult following - raised at too high a valuation in the peak era - talented teams - still founder led by strong product thinkers
Salesforce did this with Quip, Slack, etc.
toomuchtodo
It's not a bad play, like PE rollups but arguably with more potential for success from a "founder/startup friendly" perspective. PE, historically, is very good at stripping/optimizing/etc (think Bending Spoons), but I think there is magic in startups PE simply does not have that allows for a different, yet more successful, outcome (growing vs cutting your way to financial success).
We should see more of this as large, profitable startups have grown into long term private companies with no need to go public.
chii
> large, profitable startups have grown into long term private companies with no need to go public
which i think is a real problem - it prevents "mom and dad" investors from partaking in the wealth creation process, as they are not sophisticated investors and thus barred from being able to invest in these PE investments.
Public listing has become a cashing out operation, rather than a fund-raising operation, if this continues to happen more and more. And the public becomes the bag holders.
null
orliesaurus
Superhuman, the most super-email client experience, that only people in the bay area (and some folks in NYC) actually use.
How much did they pay for this? I hope not much.
iagooar
You would need close to 90k customers to get $35M revenue as they claim.
90 thousand customers sounds like a whole lot of users to me.
I use it myself and it is by far the best email experience ever created. Is it worth the money? That depends on your needs and work, I guess. CEOs laugh at the cost. Developers might think the price is nuts.
pickledoyster
Those are some of the best customers to have in case your business is under threat and you might need a bailout (M&A) in the future.
null
shortformblog
I, for one, think it’s hilarious that a company that put so much energy into being ultra-exclusive ended up getting acquired by a company with such a mainstream reputation. Grammarly’s target audience is the people who couldn’t join Superhuman in a timely fashion.
Superhuman made one of my accounts wait for four years for an invite.
Superhuman user and former Grammarly user here.
I'm a big fan of Superhuman as an email client and happily pay the premium price for it. I really hope they don't change what makes it great.
I used to love Grammarly until they essentially ruined their product - much like Dropbox did. They took an app that worked perfectly and deprecated it, replacing it with an invasive keyboard replacement that was supposed to work everywhere but performed poorly across most programs and included functionality I wasn’t interested in that is kept nagging me to use. When I complained about the issues, instead of addressing my concerns, they sent form letter responses about their commitment to privacy rather than fixing their intrusive software.
This reminds me exactly of Dropbox's transformation from simple, reliable file storage into bloated software that cluttered my computer with pop-ups and background processes. When users complained, their team never seemed to understand why we were frustrated. Then they started acquiring other services I eventually cancelled as they tried to integrate them into their core service. I eventually moved to iCloud and never looked back.
I hope Superhuman keeps their current excellent email client that I gladly pay for, rather than replacing it with some "next generation" product that nobody asked for and that would likely be inferior to what we have now. I genuinely love Superhuman as it is.