I'm not sure why this announcement has generated so much irritation in the comments-- Cloudflare has been transitioning from "DDoS protection" to "AWS competitor" for many years now, and this is just their alternative to AWS SES.
It's an email sender that you can access through an API, or directly through Workers. For those who haven't been keeping up over the years, Workers is their product for running code on Cloudflare's platform directly (an AWS Lambda competitor, more or less) and they've been trying to make it the centerpiece of an ecosystem where you deploy your code to their platform and get access to a variety of tools: databases, storage, streaming, AI, and now email sending. All of this is stuff that AWS has had for years, but some people like Cloudflare more (I certainly do).
One thing that surprised me is the price-- Cloudflare's cloud offerings are usually much cheaper, and I've saved plenty of money by migrating from AWS S3 to Cloudflare's R2. This new offering is 3x the AWS price, though. Weird. Anyway, most small companies don't send enough email for it to matter.
But getting back to the consensus in the comments here: I'm not sure why people think that they'll be worse about policing spam than AWS SES, Azure Email, etc.
I'm not sure if it's a correct impression but my impression is still that AWS is the "devil you know" and Cloudflare is less predictable with more individual decision making from high ups.
I guess they got that reputation years ago when the founders (?) got into public spats about what they would and wouldn't host. AWS is more lawyers and committees and seems more anonymous, so people don't necessarily like it more but they do trust it to be what it looks like more.
> But getting back to the consensus in the comments here: I'm not sure why people think that they'll be worse about policing spam than AWS SES, Azure Email, etc.
Cloudflare is (in)famous for not acting against spammers, fraud, piracy and other less savory groups that are hosting their stuff at/behind Cloudflare, so reasonably, people who've been affected by that are now afraid the same thing will happen with email.
We have reserved IPs for Email Service and will be protecting the reputation and fighting spam from originating on Email Service.
If we did not do so, our IPs would get flagged and then emails end up in spam or not delivered. That defeats the purpose of having a transactional Email Service. We're well aware of this.
When it comes to email delivery, you can't ignore spam. It's the bane of existence of every email sending service and the number one business challenge in that segment. After all, orchestrating delivery over SMTP is not rocket science. But getting that email to not be rejected totally IS rocket science and it's simultaneously an art form known only to a handful of email nerds working at the core of the big email sending services...
Ok, but what about as a CDN/website-proxy/WAF? I know we don't have the same automated reputation-propagation as with email, but same thing supposedly happens there, where eventually you get turned off if you don't act on lawful requests, which is exactly why Cloudflare is unavailable in Spain during La Liga matches, because Cloudflare don't take piracy streams down.
In theory, Cloudflare should take those down, when requested by legal means, but that doesn't matter. How sure are we that they'll act differently for email, instead of trying to get rid of the reputation system instead?
> getting that email to not be rejected totally IS rocket science and it's simultaneously an art form known only to a handful of email nerds working at the core of the big email sending services
It really isn't, you need a clean IP and a clean domain, send handful of emails and you're pretty much whitelisted on most services out there. Maybe you'd say I'm one of the handful, but I personally know more than a handful others who also run their own email services, just like me, and besides the usual hassle of running your own service, as long as you don't spam, your emails will arrive as usual.
Almost every SaaS (Spam as a Service) API ends up arguing its minority of legitimate users are a justified excuse for the majority of nuisance traffic.
Most cloud IP blocks already have very poor reputations, and or already on Spamhaus blacklists.
My experience has been the opposite of what you're saying: AWS SES (one of AWS's flagship products, and probably the biggest email sender in the world) is a pretty responsible anti-spam citizen. Spamhaus even wrote this article[1] praising SES's anti-spam efforts. From the article: "Amazon SES has a long-standing relationship with Spamhaus, working closely to prevent suspicious IPs and domains from impacting their network." Though I'm sure that new incidents come up daily, Spamhaus themselves seem to disagree with the notion that SES's IP blocks have "poor reputations."
I also kind of rolled my eyes at the blog post and its obsessive focus on "agents" -- definitely feels like a solution looking for a problem. But the email-sending product being promoted is probably ok, right? They just happened to write a lot of words observing that ChatGPT can, in fact, call sendmail() through their platform (if you give it access) -- a fact that shouldn't surprise anyone.
A classic "the tragedy of the commons" with the SMTP protocol.
When the cost of spamming is near 0.00, all open platforms will be abused to the tilt. We have seen the email channel get less and less reliable with our own clients (password recovery, notifications and etc).
This might evolve into a couple of oligopolies (Microsoft 365 Outlook, Google Gmail, may be some legacy email providers like Yahoo) and if you want delivery you'd need to pay them, because they'd be the verifiers that you're not a spammer.
And these platforms will have a hell of time to fight the spammers that will create millions of email addresses and spam trough them.
I don't think the protocol is necessarily the problem. For example we don't say the HTTP protocol is the problem when spammers abuse website comment forms or forums, we say it's the server on the other side.
I think the answer is somewhat the same as where we've gone with many HTTP servers: proof of work. Just like Captcha and more recently Cloudflare turnstile required you complete a task before you'd be able to access as website, senders should be required to complete a task before you'll accept their email.
It can even be a sliding scale: the higher you want the chances of the recipient seeing it to be, the more work you need to do.
However this also break emails considered "legitimate" by businesses, like marketing newsletters and other nonsense, which is why it'll likely never happen.
I've gotten my email routed to spam even though it never left the Google cloud. They don't say, "Gosh, this is coming from inside the house. Therefore it's trustworthy." Nope. The push legit mail from other Google hosted domains into spam without a second thought.
I'd be happy if we at least started punishing the large, well known and established companies for spamming us...
...you know the one, where you have email preferences, and you only have "new messages" and "commercial offers" in the settings, and you uncheck the "commercial offers" and think you're sae. Then you get a spam email from them... check the preferences again, and there's a "new product notification" preference, checked by default, and you uncheck that too. Bam! another spam! "personalized offers" option appeared, check by default. "limited time offers". "value deals", etc.
It's funny. All the examples they show in the blogpost are just things that were already pretty easy without agents. Sending an email when the CI pipeline passes, when a support request is incoming, when an order is shipped. I think we haven't found a problem for this solution.
I feel like a lot of folks down here are focusing too much on the agent part. That's purely marketing. No one who worked on the service, I am sure, was building exclusively for agent usage.
This is simply the framing device that all marketing needs to present these days.
Sending and receiving is in my mind the easy part. The hardest part is to make this work with actual AI agents. This is the same problem as with sub-agent communication because you need to implement all kinds of additional fictionality to ensure the agent is not just responding for no good reason, go into loops, etc.
More spam at scale. I wish recipients of email had more control over the conditions to which the email is delivered to them, rather than after the fact curation…
Along these lines, I am experimenting with a two-tiered system where I use Proton's sieve filters to create a sender/cc/bcc whitelist for personal emails (which alert my phone) and a non-Proton collection of burnable aliases for everything else (which do not alert my phone). It doesn't solve the problem completely, but it is mitigating it pretty well so far.
While we’re adding antiquated and shitty ways to interface with your agent, can we add fax support? Maybe direct-to-mail service for postcards and flyers?
This has been possible for many years, before agents were a thing. They will open the mail and scan the contents into a pdf for you, requires filing a form with the post office. It gets expensive because they nickel and dime you where they can. There are many more services should you wish to send snail mail.
I’ll have to take a look at this as a way to move off my homegrown serverless email on AWS. Doesn’t look like it has parity with being able to send email from many subsystems safely (with delay and veto)[1], but is pretty close on the receive side automation[2].
I seriously think this great! I’ve been saying that email is the right interface for agents for a while now. It is available anywhere, natively threaded, and works for asynchronous long-form communication. Comes with great clients as well.
I’ve been developing last three months by emailing Claude, with email threads mapping to an isolated workspace and claude -p. Works super well, especially when trying to get some coding done between everything else.
With right CLAUDE.md and a bit of workflow tooling this extends itself to building other kinds of agents as well. For example, I do my bookkeeping by emailing Claude my statements and receipts, which it then imports into a plain-text accounting system. And we’ve proven this in corporate environment as well, creating agent that can troubleshoot more complex issues by correlating diagnostic logs against product source code.
Once the basic “email agent” infrastructure is there, creating new agents becomes super simple.
Oof. I know of a startup that recently Show HN'd here, the agent mail.to, that is NOT having a good time right now. I don't know what all these new startups having moats thinner than Durex are thinking -- like, what the plan if someone does what you do, faster and cheaper?
> new startups having moats thinner than Durex are thinking
Haha, great visual. Really illustrative of what these AI startups and bootstrapped indie developers are dealing with (and, if I had to guess, why most of them don't go anywhere).
Well that part was impressive. It looks like they focused on receiving emails, that is probably even worse, as I expect OpenAI/Anthropic to add such ability directly to agents, if it really is useful.
Write an angry blog post about how big business is using their power to kill their _totally_ unique original idea that nobody could possibly copy in a hour?
Classic "is this a feature or a product?" problem. You're going to have a bad time if you spend all your effort on a feature and nothing to set it apart.
Cloudflare is very transparent about their prefixes and reverse DNS, which makes it trivial for operators who want to block the entire service, and extremely bad for Cloudflare's deliverability.
And while there are many open blacklists which I have no doubt Cloudflare monitors, there are many (including soft spam-classification signals) that are proprietary and difficult/impossible to monitor other than by watching rates of actual customer/prospect replies and engagement.
Amazon SQS has similar dynamics, and its reputation is far from stellar.
(If the Cloudflare team is reading this, and I'm missing an on-ramp to a company purchasing dedicated IPs with distinct PTR records, I do apologize! I'm not seeing documentation about this, though.)
Meanwhile, about to move all my domains(personal and business) e-mail from a provider to self-hosted using Stalwart Mail. I'll wait on the agents a while, thanks though.
There is not much developers UNDERSTAND about email, let alone believe... There's soo much to read into this product.. and it boils down to JASS ( just another spam source) .
This is a very long post just to say they're now running an SMTP server. I've been sending and receiving emails from Workers for two years; though for sending, you still need an external SMTP server like SES or Postmark.
Don't get me wrong, sending (and delivering) emails is genuinely hard. But we'll only know how good Cloudflare is at it after a couple years of real-world experience.
>Everyone already has an email address, which means everyone can already interact with your application or agent. And your agent can interact with anyone.
please no.
>Sending email that actually reaches inboxes usually means wrestling with SPF, DKIM, and DMARC records. When you add your domain to Email Service, we configure all of it automatically. Your emails are authenticated and delivered, not flagged as spam.
this is going to be an absolute nightmare for spam. i cant exactly block all of cloudflare...
it would be nice if anyone at cloudflare could write about how they plan to proactively reduce abuse of this feature, how they will respond to spam reports, what the punishment for abuse will be, etc.
I guess if they're big enough they should be working on moving off of amazon SES for emails and warming up ip addresses? Otherwise they need to keep a markup on top of amazon.
Edit: didn't realize people were paying resend $20. AWS already exists at a low price and people pick them anyway, i'm sure they're fine.
It's very easy to get started with Resend, and they have a free tier that basically works for any bootstrapped project, and by the time you have to upgrade, you can pay $20/m.
It's an email sender that you can access through an API, or directly through Workers. For those who haven't been keeping up over the years, Workers is their product for running code on Cloudflare's platform directly (an AWS Lambda competitor, more or less) and they've been trying to make it the centerpiece of an ecosystem where you deploy your code to their platform and get access to a variety of tools: databases, storage, streaming, AI, and now email sending. All of this is stuff that AWS has had for years, but some people like Cloudflare more (I certainly do).
One thing that surprised me is the price-- Cloudflare's cloud offerings are usually much cheaper, and I've saved plenty of money by migrating from AWS S3 to Cloudflare's R2. This new offering is 3x the AWS price, though. Weird. Anyway, most small companies don't send enough email for it to matter.
But getting back to the consensus in the comments here: I'm not sure why people think that they'll be worse about policing spam than AWS SES, Azure Email, etc.
I guess they got that reputation years ago when the founders (?) got into public spats about what they would and wouldn't host. AWS is more lawyers and committees and seems more anonymous, so people don't necessarily like it more but they do trust it to be what it looks like more.
Probably just a function of time and size.
Cloudflare is (in)famous for not acting against spammers, fraud, piracy and other less savory groups that are hosting their stuff at/behind Cloudflare, so reasonably, people who've been affected by that are now afraid the same thing will happen with email.
We have reserved IPs for Email Service and will be protecting the reputation and fighting spam from originating on Email Service.
If we did not do so, our IPs would get flagged and then emails end up in spam or not delivered. That defeats the purpose of having a transactional Email Service. We're well aware of this.
In theory, Cloudflare should take those down, when requested by legal means, but that doesn't matter. How sure are we that they'll act differently for email, instead of trying to get rid of the reputation system instead?
> getting that email to not be rejected totally IS rocket science and it's simultaneously an art form known only to a handful of email nerds working at the core of the big email sending services
It really isn't, you need a clean IP and a clean domain, send handful of emails and you're pretty much whitelisted on most services out there. Maybe you'd say I'm one of the handful, but I personally know more than a handful others who also run their own email services, just like me, and besides the usual hassle of running your own service, as long as you don't spam, your emails will arrive as usual.
Most cloud IP blocks already have very poor reputations, and or already on Spamhaus blacklists.
People have a right to choose to be upset. =3
[1] https://www.spamhaus.org/resource-hub/service-providers/how-...
When the cost of spamming is near 0.00, all open platforms will be abused to the tilt. We have seen the email channel get less and less reliable with our own clients (password recovery, notifications and etc).
This might evolve into a couple of oligopolies (Microsoft 365 Outlook, Google Gmail, may be some legacy email providers like Yahoo) and if you want delivery you'd need to pay them, because they'd be the verifiers that you're not a spammer.
And these platforms will have a hell of time to fight the spammers that will create millions of email addresses and spam trough them.
I think the answer is somewhat the same as where we've gone with many HTTP servers: proof of work. Just like Captcha and more recently Cloudflare turnstile required you complete a task before you'd be able to access as website, senders should be required to complete a task before you'll accept their email.
It can even be a sliding scale: the higher you want the chances of the recipient seeing it to be, the more work you need to do.
However this also break emails considered "legitimate" by businesses, like marketing newsletters and other nonsense, which is why it'll likely never happen.
...you know the one, where you have email preferences, and you only have "new messages" and "commercial offers" in the settings, and you uncheck the "commercial offers" and think you're sae. Then you get a spam email from them... check the preferences again, and there's a "new product notification" preference, checked by default, and you uncheck that too. Bam! another spam! "personalized offers" option appeared, check by default. "limited time offers". "value deals", etc.
This is simply the framing device that all marketing needs to present these days.
"Please stop talking about the thing we can't stop talking about"
Sending and receiving is in my mind the easy part. The hardest part is to make this work with actual AI agents. This is the same problem as with sub-agent communication because you need to implement all kinds of additional fictionality to ensure the agent is not just responding for no good reason, go into loops, etc.
My $0.02 from experience.
https://www.virtualpostmail.com/
[1] https://github.com/mlhpdx/email-origin [2] https://github.com/mlhpdx/email-delivery
I’ve been developing last three months by emailing Claude, with email threads mapping to an isolated workspace and claude -p. Works super well, especially when trying to get some coding done between everything else.
With right CLAUDE.md and a bit of workflow tooling this extends itself to building other kinds of agents as well. For example, I do my bookkeeping by emailing Claude my statements and receipts, which it then imports into a plain-text accounting system. And we’ve proven this in corporate environment as well, creating agent that can troubleshoot more complex issues by correlating diagnostic logs against product source code.
Once the basic “email agent” infrastructure is there, creating new agents becomes super simple.
Haha, great visual. Really illustrative of what these AI startups and bootstrapped indie developers are dealing with (and, if I had to guess, why most of them don't go anywhere).
Well that part was impressive. It looks like they focused on receiving emails, that is probably even worse, as I expect OpenAI/Anthropic to add such ability directly to agents, if it really is useful.
Cloudflare is very transparent about their prefixes and reverse DNS, which makes it trivial for operators who want to block the entire service, and extremely bad for Cloudflare's deliverability.
And while there are many open blacklists which I have no doubt Cloudflare monitors, there are many (including soft spam-classification signals) that are proprietary and difficult/impossible to monitor other than by watching rates of actual customer/prospect replies and engagement.
Amazon SQS has similar dynamics, and its reputation is far from stellar.
(If the Cloudflare team is reading this, and I'm missing an on-ramp to a company purchasing dedicated IPs with distinct PTR records, I do apologize! I'm not seeing documentation about this, though.)
$0.35 per 1,000 emails
Here are the limits:
"Your account may have daily sending limits based on Cloudflare's assessment of your account standing. "
Source: https://developers.cloudflare.com/email-service/platform/pri... https://developers.cloudflare.com/email-service/platform/lim...
Things developers believe about email
Don't get me wrong, sending (and delivering) emails is genuinely hard. But we'll only know how good Cloudflare is at it after a couple years of real-world experience.
please no.
>Sending email that actually reaches inboxes usually means wrestling with SPF, DKIM, and DMARC records. When you add your domain to Email Service, we configure all of it automatically. Your emails are authenticated and delivered, not flagged as spam.
this is going to be an absolute nightmare for spam. i cant exactly block all of cloudflare...
it would be nice if anyone at cloudflare could write about how they plan to proactively reduce abuse of this feature, how they will respond to spam reports, what the punishment for abuse will be, etc.
How's that compare?
Looks better than fixed $20 for Resend.
I like Resend, a lot, but this is probably something I can't pass up, especially if it does what it says on the tin
Edit: didn't realize people were paying resend $20. AWS already exists at a low price and people pick them anyway, i'm sure they're fine.
It would be interesting to send GDPR requests and have Cloudflare figure out all of the parties who got or use your mail.