Nitrosend
Email for AI agents. They sign up, send and reply.
185 followers
Email for AI agents. They sign up, send and reply.
185 followers
Nitrosend is full-stack email built for AI agents. Point any agent at nitrosend.com/SKILL.md and it signs itself up, onboards, connects your domain and sends. Live today: marketing and transactional email, real inboxes for agents on your own domain (beta, by request), and 1-1 customer replies with human escalation. Coming soon: personalised outreach run by your agent. In development: goal-based agentic marketing. Humans approve. Agents operate.
Free Options
Launch Team / Built With
Nitrosend
Hey Product Hunt, George here 👋🏻
Nitrosend came out of running Obooko (producthunt.com/products/obooko, #2 Product of the Day here in February). Running a free ebook platform with 840,000 readers on its list, entirely through an AI agent, convinced me the dashboard era of email is done. It's also my third email company: SmartrMail (acquired 2022) and Sendicate before it, with six billion emails sent between them.
Six weeks ago we launched Nitrosend as headless email you run from Claude or Codex instead of a UI. Today it goes further: the agent gets an inbox, and get send for itself. Point any agent at nitrosend.com/SKILL.md and it signs up, connects your domain, sorts billing and sends its first email. no human required.
The bit I'm proudest of is Agent Inboxes (beta): real addresses for your agents on the domain you own, created BY your agent, like ernest@yourcompany.com, sylvia@yourcompany.com. They send, receive and hold 1-1 conversations with your customers, and escalate to a human when it matters. Gmail actively fights this. we built the place where it's native. Because very soon this will be table stakes for your agent, so we're shipping it now.
Coming real soon: Agentic Outreach, where your agent researches prospects, writes personally and works the replies. Get a look alike audience from your current customer list, zoom in on your ICP, give them personalised messaging. No start up founder I know likes doing spray and pray outbound, but some still do it as the tools aren't good enough to really make it relevant. We're solving that.
In development: set a goal for your brand's email and Nitrosend does the rest.
Agentic signup and onboarding are shipped, headless full stack newsletters, flows, transactional email all live sending and delighting our first users (including me on Obooko!). Agent Inboxes are beta by request (ask in app and we can turn it on for you), Outreach is coming soon, goal-based is in dev.
Super excited to ship this, ask any question in the comments!
The "agent points at SKILL.md, signs itself up and connects the domain" onboarding is slick, but the thing I'd test first is deliverability once an agent is the one hitting send. When an agent ramps a fresh domain, does Nitrosend enforce its own warmup and rate limits (plus SPF/DKIM/DMARC verification) before it'll send at volume, or can an agent torch a new domain's reputation on day one? That guardrail is what would make me comfortable letting an agent operate unattended.
Nitrosend
@hazy0 right question to ask, deliverability is the bit we're most paranoid about (third email company, we've been burned before). Nothing sends until SPF, DKIM and DMARC verify on the domain, sending is quota'd so a fresh domain can't blast at volume on day one, and bounce and complaint monitoring runs on every send. The agent gets the same guardrails a good human sender would, it just actually follows them.
Congrats @gthartley for the launch 🙌 love the target audience lookalike concept for outbound, qq how do you handle privacy and security for AI-managed conversations?
Nitrosend
@priya_kushwaha1 Thanks Priya! On privacy: the agent only sees what you scope it to. Access is per-brand via MCP or API keys, your contact data never trains anything, and sends have approval gates by default so nothing goes out to a list without a human saying so. we've even shipped an optional safe mode if people don't want their agent to be able to do any bulk sends (requires a human clicking final send button)
Love the idea - i) Is it visible on the receiving end that its agentic? ii) i it EU compliant i.e. opt out options per default ?
Nitrosend
@nikolaos_chr i) Nope, nothing on the receiving end flags it as agentic. It's a standard email sent from whatever address you set up on your own domain, same as if you'd written it in Gmail. ii) Yep, GDPR compliant, and unsubscribe and opt-out is on by default.
agent inboxes that escalate 'when it matters' is the real product. deciding when to hand off is the hard part. what signal makes it escalate vs reply itself?
Nitrosend
@andrewzakonov spot on, the handoff judgment is the hard part. You scope what the agent owns, and it errs on the side of escalating: money, cancellations, legal, an upset human, or anything ambiguous gets handed to you, the routine stuff it answers. That tuning is exactly why inboxes are beta by request right now, we're calibrating the judgment with real traffic before opening it wide.
kinda wild concept. whats stopping this from turning inboxes into agent spam over time?
Nitrosend
@doganakbulut the same thing that stops human spam: consent and reputation. Agents on Nitrosend send to lists people opted into, unsubscribe is on by default, and bounce and complaint monitoring pulls things up fast. Spam torches the sender's own domain first, so the incentives point the right way. Honestly the agents are better behaved than most humans with a CSV export 😄
Congrats on the launch! I'm wondering if companies would actually use AI agents to send mails because for me it's not about the technology but more around trust that there should be a human verifying. Have you seen customers with similar perspectives ?
Nitrosend
@reda_roqai_chaoui thanks Reda! And yes, constantly. That's why approval gates are on by default: the agent drafts and preps, a human clicks send on anything bulk, and there's a draft-only safe mode for teams that want zero autonomy. Trust ramps the way it did with cruise control, you supervise first, hand over more as it earns it. Plenty of our users start fully supervised and loosen up within a week or two.