<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[Numan Hamza]]></title><description><![CDATA[Numan Hamza]]></description><link>https://numanhamza.hashnode.dev</link><generator>RSS for Node</generator><lastBuildDate>Sat, 21 Nov 2026 14:54:12 GMT</lastBuildDate><atom:link href="https://numanhamza.hashnode.dev/rss.xml" rel="self" type="application/rss+xml"/><language><![CDATA[en]]></language><ttl>60</ttl><item><title><![CDATA[How WarmySender's self-hosted cold email engine keeps send limits in the code]]></title><description><![CDATA[I just open-sourced the cold email engine behind WarmySender. It is a self-hosted Instantly alternative: you run it with docker compose up, connect your…]]></description><link>https://numanhamza.hashnode.dev/how-warmysender-s-self-hosted-cold-email-engine-keeps-send-limits-in-the-code</link><guid isPermaLink="true">https://numanhamza.hashnode.dev/how-warmysender-s-self-hosted-cold-email-engine-keeps-send-limits-in-the-code</guid><category><![CDATA[Open Source]]></category><category><![CDATA[#selfhosted]]></category><category><![CDATA[TypeScript]]></category><category><![CDATA[email]]></category><dc:creator><![CDATA[Numan Hamza]]></dc:creator><pubDate>Fri, 20 Nov 2026 19:51:31 GMT</pubDate><content:encoded><![CDATA[<p>I just open-sourced the cold email engine behind WarmySender. It is a self-hosted Instantly alternative: you run it with <code>docker compose up</code>, connect your own mailboxes, and the app sends from those mailboxes on your server.</p>
<p>The part I want to walk through here is not the UI. It is how sending actually works, and why an AI agent connected over MCP cannot turn it into a blast tool.</p>
<p>The constants, the queue, and the agent lock all live in the public repo. I am describing v0.1.0 as it is today.</p>
<h2>What you get when you start it</h2>
<pre><code class="language-bash">git clone https://github.com/warmysender-oss/warmysender.git
cd warmysender
docker compose up
</code></pre>
<p>Open <a href="http://localhost:3000">http://localhost:3000</a>. The first account is the admin. Compose brings up the app, the worker, and PostgreSQL. Jobs stay in Postgres through <a href="https://github.com/timgit/pg-boss">pg-boss</a>.</p>
<p>On a laptop, development defaults are fine. <code>NODE_ENV=production</code> refuses those defaults, so you have to set <code>ENCRYPTION_KEY</code> and <code>SESSION_SECRET</code> before you expose the server.</p>
<h2>The worker is a loop, not a firehose</h2>
<p>The worker (<code>apps/worker/src/index.ts</code>) starts a <code>PgBoss</code> client on the same Postgres database, then creates three queues:</p>
<table>
<thead>
<tr>
<th>Queue</th>
<th>Cron</th>
<th>What it does</th>
</tr>
</thead>
<tbody><tr>
<td><code>heartbeat</code></td>
<td><code>* * * * *</code> (every minute)</td>
<td>Writes a heartbeat so you can see the worker is alive</td>
</tr>
<tr>
<td>inbox poll</td>
<td><code>*/5 * * * *</code> (every five minutes)</td>
<td>Polls IMAP, files replies and bounces</td>
</tr>
<tr>
<td><code>send-tick</code></td>
<td><code>* * * * *</code> (every minute)</td>
<td>Asks the scheduler who is allowed to send right now</td>
</tr>
</tbody></table>
<p>The important one is <code>send-tick</code>. Every minute the worker runs <code>executeSendTick</code>. That function does not "send the campaign." It asks the scheduler whether this mailbox, at this moment, is allowed to send one email.</p>
<p>A tick can fail. When it does, the worker logs it and rethrows, so a broken send cycle is visible in the logs instead of disappearing.</p>
<p>Inbox mail is on a slower loop on purpose. Replies and bounces matter, but they do not need to be checked every minute.</p>
<h2>Caps are constants, then they are enforced again at send time</h2>
<p>The numbers are not only in the README. They are named constants in <code>packages/core/src/scheduler/constants.ts</code>:</p>
<ul>
<li><code>DEFAULT_DAILY_CAP = 30</code></li>
<li><code>HARD_MAX_DAILY_CAP = 50</code></li>
<li><code>RAMP_CAP_DAYS_1_TO_7 = 10</code></li>
<li><code>RAMP_CAP_DAYS_8_TO_14 = 20</code></li>
<li>Default random gap: 4 to 9 minutes</li>
<li>Transient SMTP retries: 3 attempts, with 5-minute then 15-minute backoff</li>
<li>Three sign-in failures in a row pause the mailbox</li>
<li>After an auth failure, the worker waits 30 minutes before trying that login again</li>
</ul>
<p><code>packages/core/src/scheduler/caps.ts</code> turns those into the cap for today:</p>
<ol>
<li><code>clampDailyCap</code> floors the configured cap and never returns more than 50. Zero or negative means the mailbox cannot send.</li>
<li><code>rampDayNumber</code> counts calendar days from <code>rampStartedOn</code> in the mailbox's local date. If the clock is skewed and "today" is somehow before the start date, it stays on day 1 so the ramp cannot be skipped.</li>
<li><code>effectiveDailyCap</code> takes the lower of the clamped mailbox cap and the ramp cap. Days 1 to 7 are 10. Days 8 to 14 are 20. After that, the mailbox's own cap applies, still clamped at 50.</li>
</ol>
<p>The form validation matches this. <code>validateDailyCap</code> accepts 30 and 50, and rejects 51, 0, fractions, and junk strings. You cannot type 200 into the mailbox form and have it stick.</p>
<h2>A send has to pass a checklist</h2>
<p><code>packages/core/src/scheduler/evaluate.ts</code> is the gate. Before a mailbox sends, the scheduler checks, in order:</p>
<ul>
<li>The schedule itself is valid (window and gap values make sense)</li>
<li>Today is a sending day for that mailbox</li>
<li>The current time, in the mailbox's timezone, is inside its sending window</li>
<li>The ramp day can be computed</li>
<li>Today's effective cap is known, and the mailbox is still under it</li>
<li>The random gap since the last send has elapsed</li>
</ul>
<p>Fail any one of those and that tick does not send. The reasons have names (<code>outside_window</code>, <code>gap_not_elapsed</code>, <code>invalid_schedule</code>, and so on), which is what you want when you are debugging a quiet mailbox at 11pm.</p>
<p>One email at a time per mailbox is also a rule, not a preference. The gap (default 4 to 9 minutes) is why a tick every minute does not mean a send every minute.</p>
<h2>Where agents are locked out</h2>
<p>The app ships a REST API (<code>/api/v1</code>) and an MCP server (<code>/mcp</code>) for Claude, ChatGPT, Cursor, Codex, OpenClaw, Hermes Agent, and other MCP clients. Agents can build a campaign. They cannot send a message themselves, and they cannot raise a limit.</p>
<p>The MCP instructions say so in plain words: "Tools never send directly and can never raise a limit; the scheduler sends inside safe limits."</p>
<p>The tools in <code>apps/server/src/mcp/tools.ts</code> include things like <code>create_contact</code>, <code>create_campaign</code>, <code>add_campaign_step</code>, <code>start_campaign</code>, <code>pause_campaign</code>, <code>enroll</code>, <code>list_replies</code>, and <code>draft_reply</code>. There is a <code>draft_reply</code> tool. There is not a "send this reply now" tool. A draft stays a draft until a person sends it from the inbox.</p>
<p>The API also strips a class of fields on the way in. <code>apps/server/src/api/limits.ts</code> walks every nested object in a request and rejects keys such as <code>daily_cap</code>, <code>min_gap_minutes</code>, <code>window_start</code>, and <code>ramp</code>, no matter the capitalization or hyphens. The response is <code>403</code> with <code>limit_locked</code> and the message: "An agent can never raise a limit. Daily cap, ramp, sending window, and the gap between sends are fixed in code."</p>
<p>That is the design I wanted in the open version. The scary part of handing a campaign to an agent is not that it can write a sequence. It is that it might try to send 2,000 emails overnight. In this codebase, that request should fail before the worker ever sees it.</p>
<h2>What v0.1 does not do</h2>
<p>I want this to be accurate, because v0.1 is the first public release.</p>
<ul>
<li>It sends the message as you wrote it. There is no open or click tracking.</li>
<li>Warmup, email verification, the leads database, LinkedIn, and Instagram stay in WarmySender Cloud. The open-source app can optionally use warmup and verification through a WarmySender API key. The core app runs with no key.</li>
<li>Microsoft 365 / Outlook needs your own Entra app, documented in the repo.</li>
</ul>
<p>If you try it, the repo and GitHub Discussions are the right place to tell me what breaks: <a href="https://github.com/warmysender-oss/warmysender">https://github.com/warmysender-oss/warmysender</a></p>
]]></content:encoded></item><item><title><![CDATA[How to receive email in your app: MX record, webhook, reply in thread]]></title><description><![CDATA[Inbound email without running a mail server: the DNS record, a webhook handler that verifies its sender, reading the reply without the quoted history, and…]]></description><link>https://numanhamza.hashnode.dev/how-to-receive-email-in-your-app-mx-record-webhook-reply-in-thread</link><guid isPermaLink="true">https://numanhamza.hashnode.dev/how-to-receive-email-in-your-app-mx-record-webhook-reply-in-thread</guid><category><![CDATA[email]]></category><category><![CDATA[webhooks]]></category><category><![CDATA[Node.js]]></category><category><![CDATA[api]]></category><dc:creator><![CDATA[Numan Hamza]]></dc:creator><pubDate>Fri, 20 Nov 2026 14:36:45 GMT</pubDate><content:encoded><![CDATA[<p><em>By the AgentiSend team. The examples use our inbound API, which we can test in CI. The steps (MX record, webhook,
fetch, reply with threading headers) are the same with any provider that receives mail. This article was written with the assistance of an AI writing tool and checked by the AgentiSend team, who stand behind every claim in it.</em></p>
<p><strong>Short answer:</strong> publish one MX record on a subdomain pointing at an inbound email API, subscribe to its "email
received" webhook, verify the signature on the raw body, read the message (use the version with quoted history
stripped), and reply through the API so the customer's mail client threads it. Skip auto-replies to auto-replies. The
handler below does all of it and runs in our CI.</p>
<p>Sending email from an app is a solved problem. Receiving it is where teams still run Postfix on a VM or poll an IMAP
inbox every minute. Both work, and both leave you parsing MIME, stripping quoted replies and guessing at threading by
hand.</p>
<p>An inbound email API does the mail-server part for you: you point an MX record at it, it accepts the message, stores
it, and calls your webhook. Your code deals with JSON. This post walks through the whole path, from the DNS record to a
threaded reply, with a handler our CI runs on every change.</p>
<p>Use cases where this pays off:</p>
<ul>
<li><strong>Support inboxes</strong> where replies should land in your app, not in a shared mailbox.</li>
<li><strong>Reply-by-email</strong> on notifications ("reply to this email to comment").</li>
<li><strong>AI agents that read and answer mail</strong>, which need the reply text without three pages of quoted history.</li>
<li><strong>Document intake</strong>: invoices or statements forwarded to an address, attachments extracted.</li>
</ul>
<h2>Step 1: give mail somewhere to go</h2>
<p>Mail for a domain goes to the host in its MX record. To receive at <code>support@inbound.acme.com</code>, you turn receiving on for
that domain and publish one record:</p>
<pre><code class="language-text">inbound.acme.com.  MX  10  inbound.agentisend-dns.com.
</code></pre>
<p>Two rules save you a bad afternoon:</p>
<ol>
<li><strong>Use a subdomain if the apex already receives mail.</strong> If <code>acme.com</code> is on Google Workspace or Microsoft 365, adding
a second MX there splits your company mail. <code>inbound.acme.com</code> or <code>reply.acme.com</code> keeps them apart.</li>
<li><strong>Make it the lowest priority number on that name</strong>, so it is the host senders try first.</li>
</ol>
<p>Mail to a domain that is not verified, or has receiving off, is refused at the door and never stored. Messages up to
10 MB are accepted, for up to 20 recipients on your domain per message.</p>
<p>You can also decide who may write to the domain. The default policy stores every sender a block rule does not match;
an allow-list policy stores only senders that match an allow rule. A block match wins over an allow match, and a
refused message is not stored and gets no reply, so a block list does not turn your domain into a bounce generator.</p>
<h2>Step 2: subscribe to <code>email.received</code></h2>
<p>Every stored message fires an <code>email.received</code> webhook. The payload names the message (<code>email_id</code>), the sender, the
recipients, the subject, the <code>Message-ID</code>, the attachment names and <code>extracted_text</code>: the new part of the message with
quoted history, signature and "sent from my phone" lines removed. The body, headers and attachment bytes are read
through the API, so a webhook log never holds whole messages.</p>
<h2>What the webhook carries</h2>
<p>An <code>email.received</code> delivery, trimmed to the fields a handler uses (values from our test):</p>
<pre><code class="language-json">{
  "type": "email.received",
  "data": {
    "email_id": "0b6e…",
    "from": "bob@partner.test",
    "to": ["support@acme.com"],
    "received_for": ["support@acme.com"],
    "subject": "Where is my order?",
    "message_id": "&lt;q1@partner.test&gt;",
    "extracted_text": "Order 1042 has not arrived.",
    "attachments": [],
    "inbound_verdict": { "verdict": "clear", "reasons": [] }
  }
}
</code></pre>
<p>The full body, headers and attachment bytes are not in the webhook; they are one authenticated read away. That keeps
message contents out of your webhook logs.</p>
<p><strong>Delivery and retries.</strong> A delivery that does not get a 2xx is attempted up to 10 times: once, then after 5 seconds,
5 minutes, 30 minutes, 2 hours, 5 hours, 10 hours, 10 hours, 24 hours and 48 hours, then dead-lettered (kept, and
replayable from the console). Delivery is at least once, so the same event id can arrive twice. Answer 2xx as soon as
you have stored the event and do slow work after; the idempotency key on the reply covers a duplicate.</p>
<h2>Step 3: a handler that verifies, reads and replies</h2>
<p>The handler does five things: verify the signature on the raw body, ignore events it does not handle, read the full
message, skip automatic mail, reply in thread. It uses the standard <code>Request</code> and <code>Response</code> types, so it drops into a Next.js route
handler, Hono, a Cloudflare Worker or Bun unchanged.</p>
<pre><code class="language-ts">import { AgentiSend } from 'agentisend';
import { SIGNATURE_HEADER, verifyWebhookSignature } from 'agentisend/webhook-signing';

const agentisend = new AgentiSend(process.env.AGENTISEND_API_KEY);

export async function handleWebhook(request: Request): Promise&lt;Response&gt; {
  // Verify the bytes that arrived, before any JSON parser touches them.
  const rawBody = await request.text();
  const signature = request.headers.get(SIGNATURE_HEADER) ?? '';
  if (!verifyWebhookSignature(process.env.AGENTISEND_WEBHOOK_SECRET!, signature, rawBody)) {
    return new Response('bad signature', { status: 401 });
  }

  const event = JSON.parse(rawBody);
  if (event.type !== 'email.received') return new Response(null, { status: 204 });

  // The event names the sender and subject; the whole message is one read away.
  const message = await agentisend.receiving.get(event.data.email_id);

  // Never auto-reply to an auto-reply or a mailing list (RFC 3834), or two systems answer each other forever.
  const autoSubmitted = message.headers?.['auto-submitted']?.toLowerCase();
  if ((autoSubmitted &amp;&amp; autoSubmitted !== 'no') || message.headers?.['list-id']) {
    return new Response(null, { status: 204 });
  }

  // extracted_text is the reply without its quoted history or signature.
  const said = message.extracted_text ?? message.text ?? '';

  await agentisend.receiving.reply(
    message.id,
    { text: `Thanks, we got your message:\n\n&gt; ${said.split('\n').join('\n&gt; ')}\n\nA person replies today.` },
    { idempotencyKey: `ack-${message.id}` },
  );
  return new Response(null, { status: 204 });
}
</code></pre>
<p>What each part is doing:</p>
<p><strong>Signature first, on the raw text.</strong> The signature covers the exact bytes sent. A framework that parses JSON and
re-serialises it changes whitespace and key order, and the check fails, or worse, you skip the check because it
"keeps failing". Read the body as text, verify, then parse. Our test sends the handler a forged signature and expects
401, then a correctly signed real delivery and expects 204.</p>
<p><strong><code>extracted_text</code> before <code>text</code>.</strong> A customer reply usually carries the whole previous conversation below it. In the
test, the incoming message is one line ("Order 1042 has not arrived.") followed by "On Mon, Acme wrote:" and the
quoted earlier email. The acknowledgement quotes only the new line; the old quote never comes back. <code>extracted_text</code>
is null when it could not be produced, so fall back to <code>text</code>, then <code>html</code>.</p>
<p><strong><code>receiving.reply</code> threads for you.</strong> The reply goes to the original sender and carries <code>In-Reply-To</code> and
<code>References</code> taken from the stored message, so Gmail and Outlook file it under the same conversation. Budgets, the kill
switch and suppressions apply to the reply like any send.</p>
<p><strong>No reply to automatic mail.</strong> RFC 3834 says automatic responses should not be sent to a message whose
<code>Auto-Submitted</code> header has any value other than "no". Out-of-office replies and other auto-responders set it; mailing
lists set <code>List-Id</code>. Without this check, your acknowledgement and someone's out-of-office answer each other until a
limit stops them. In our test an <code>Auto-Submitted: auto-replied</code> message is stored and gets no reply.</p>
<p><strong>An idempotency key per message.</strong> Webhooks are delivered at least once. If your handler times out after replying,
the retry would reply again. <code>ack-${message.id}</code> makes the second reply return the first one's id instead.</p>
<h2>Step 4: attachments and suspect mail</h2>
<p>Attachments come back from <code>GET /emails/receiving/{id}/attachments</code> with a <code>download_url</code> that needs no API key and
expires after an hour, which suits handing a file to a worker without passing credentials around. For PDFs, Word and
Excel files, CSV, Markdown, JSON and HTML there is also a <code>text_url</code> with the extracted text, so an invoice-intake job
does not need its own PDF parser. Files always download as plain bytes, whatever type the sender declared.</p>
<p>If an AI agent reads the inbox, treat every message body as data, not instructions. Each received message carries a
verdict: <code>clear</code>, <code>review</code> or <code>unavailable</code>. <code>review</code> means the message tries to instruct an assistant, hides text,
carries a long encoded block or asks for an action from a Reply-To that differs from the From address. A domain in
hold mode stores <code>review</code> messages as held and does not fire <code>email.received</code> until a person releases them. The
verdict is a label, not a guarantee; keep the agent's tools narrow either way.</p>
<p>One more boundary: the key you give an agent to <em>send</em> cannot read received mail. Reading the inbox takes a
full-access key, because a sending key handed to an agent would otherwise read password resets and invoices strangers
send to the domain. Replying stays a send, so a sending key that was handed a message id by your webhook can still
answer it.</p>
<h2>Limits at a glance</h2>
<table>
<thead>
<tr>
<th>Limit</th>
<th>Value</th>
</tr>
</thead>
<tbody><tr>
<td>Message size</td>
<td>10 MB</td>
</tr>
<tr>
<td>Recipients on your domain per message</td>
<td>20</td>
</tr>
<tr>
<td>Allow and block rules per domain</td>
<td>1,000</td>
</tr>
<tr>
<td>Attachment download links</td>
<td>expire after 1 hour</td>
</tr>
<tr>
<td>Extracted attachment text</td>
<td>PDF, .docx, .xlsx, CSV, text, Markdown, JSON, HTML; files up to 10 MB; first 250,000 characters</td>
</tr>
<tr>
<td>Webhook delivery attempts</td>
<td>10, over about four days, then dead-lettered</td>
</tr>
</tbody></table>
<h2>Replying to mail that arrived somewhere else</h2>
<p>If the inbox lives in a help desk or Gmail and only the reply goes through the API, copy the threading headers
yourself: <code>In-Reply-To</code> is the <code>Message-ID</code> of the message you answer, and <code>References</code> is that message's
<code>References</code> followed by its <code>Message-ID</code>, each in angle brackets, separated by spaces. Get those two right and the
customer's client threads the reply, whichever system sent it. The <a href="https://agentisend.com/docs/guides/inbound">inbound guide</a>
has the length limits for both headers.</p>
<h2>Testing without real mail</h2>
<p>You do not need a domain or DNS to test the handler. In our CI the test stores a raw RFC 5322 message directly, takes
the real <code>email.received</code> delivery the API produced, signs it with the same function production uses, and calls the
handler. That covers three common inbound bugs: verifying a parsed body, replying with the
quoted history, and replying twice on a webhook retry.</p>
<h2>FAQ</h2>
<p><strong>What DNS change do I need?</strong> One MX record on the name you receive at, pointing at the provider's inbound host
(<code>inbound.agentisend-dns.com</code> here), with the lowest priority number on that name. Use a subdomain if your main
domain already receives company mail.</p>
<p><strong>Webhook or polling?</strong> Webhook. Polling <code>GET /emails/receiving</code> works for a batch job, but the webhook arrives as the
message is stored and carries enough to route it without a read.</p>
<p><strong>How are attachments handled?</strong> Listed in the webhook by name and type; fetched through the API with a one-hour
download link, plus extracted text for common document types.</p>
<p><strong>Can my app reply in the same thread?</strong> Yes: the reply call sets <code>In-Reply-To</code> and <code>References</code> from the stored
message. For mail received elsewhere, set those two headers yourself.</p>
<p><strong>Do I still need an email parser?</strong> Not for MIME, quoted-reply removal or attachments; the API returns those parsed.</p>
<p><strong>Can an AI agent read the inbox safely?</strong> Treat bodies as data. Each message carries a <code>clear</code> / <code>review</code> /
<code>unavailable</code> verdict, a domain can hold <code>review</code> mail until a person releases it, and a send-only key cannot read the
inbox at all.</p>
<h2>Other options</h2>
<ul>
<li><strong>Self-hosted:</strong> Postfix or Haraka in front of your app, or <a href="https://github.com/postalserver/postal">Postal</a>, an open-source
platform that receives as well as sends. You own MIME parsing, quote stripping and spam handling.</li>
<li><strong>Other APIs:</strong> Mailgun calls its inbound feature Routes; Resend also receives mail. Compare payload shape, whether
the quoted reply is stripped for you, whether a sending key can read the inbox, and attachment handling.</li>
</ul>
<p>The decision rule we use: if fewer than a few hundred messages a day arrive and nobody needs them in the app, a shared
mailbox is fine. Once replies have to drive something in your product, or an agent has to read them, use an API and
let the webhook be the only thing your code listens to.</p>
<hr />
<p><em>The AgentiSend team builds a transactional email API for applications and AI agents. The handler above is run
against our API on every change; the assertions quoted in this post are from that test.</em></p>
]]></content:encoded></item><item><title><![CDATA[An Instantly and Lemwarm alternative with warmup, cold email and LinkedIn in one tool]]></title><description><![CDATA[Lemwarm warms mailboxes. Instantly sends cold email and warms them too.]]></description><link>https://numanhamza.hashnode.dev/an-instantly-and-lemwarm-alternative-with-warmup-cold-email-and-linkedin-in-one-tool</link><guid isPermaLink="true">https://numanhamza.hashnode.dev/an-instantly-and-lemwarm-alternative-with-warmup-cold-email-and-linkedin-in-one-tool</guid><category><![CDATA[email]]></category><category><![CDATA[email marketing]]></category><category><![CDATA[sales]]></category><category><![CDATA[SaaS]]></category><category><![CDATA[marketing]]></category><dc:creator><![CDATA[Numan Hamza]]></dc:creator><pubDate>Mon, 09 Nov 2026 18:09:58 GMT</pubDate><content:encoded><![CDATA[<p><strong>WarmySender is an email warmup and cold email platform, and an Instantly and Lemwarm alternative, that also runs LinkedIn outreach, verifies email addresses and can be driven by an AI agent that can't raise any sending limit.</strong> Peer-to-peer warmup is included on every paid plan, and replies from every channel land in one inbox.</p>
<p>I'm Numan Hamza, the founder of WarmySender. Instantly and Lemwarm are the two names I hear most when people describe their current setup, so this post compares all three. WarmySender facts come from <a href="https://warmysender.com">warmysender.com</a> and its <a href="https://warmysender.com/llms.txt">llms.txt</a>. Instantly facts come from <a href="https://help.instantly.ai/en/articles/10273259-instantly-plans-overview">Instantly's help center</a>, and Lemwarm facts from <a href="https://www.lemwarm.com">lemwarm.com</a>, as of November 2026.</p>
<h2>First, what each tool is</h2>
<p>These three tools aren't quite the same kind of product, and that's the most useful thing to know before comparing them.</p>
<ul>
<li><strong>Lemwarm</strong> is a deliverability tool. It warms up your inbox across a network of 20,000+ real domains, checks your technical setup and gives you a deliverability score. It doesn't send your campaigns. It's sold per mailbox on its own, and it's included free with every lemlist plan.</li>
<li><strong>Instantly</strong> is a cold email platform. Its Outbound plans let you connect and warm up unlimited email accounts, upload leads and send campaigns. Lead data comes from separate Credits plans, and Instantly has no warmup-only plan.</li>
<li><strong>WarmySender</strong> covers warmup and cold email together, and adds LinkedIn outreach, real-time email verification, a 200M+ B2B lead database and a unified inbox. Instagram and WhatsApp outreach are available as per-seat add-ons.</li>
</ul>
<h2>Why I built it</h2>
<p>The setup I kept seeing was three or four subscriptions: one tool to warm the mailboxes, another to send, another for LinkedIn and another to verify the list. Each had its own limits and dashboard, and none knew what the others were doing.</p>
<p>So WarmySender puts them in one place, with one scheduler pacing everything:</p>
<ul>
<li><strong>Peer-to-peer warmup.</strong> Mailboxes in the network exchange real conversations with other verified mailboxes, with multi-turn threads. The network exchanges 900K+ warmup emails a month, and 8,500+ mailboxes have been warmed to date. It works with Gmail, Outlook and any SMTP provider.</li>
<li><strong>Peer protection.</strong> If a mailbox's spam rate climbs to about 5% over the last 7 days, its daily warmup volume is cut back while it recovers. Above about 15%, it's taken out of the pool, which protects everyone else in the network.</li>
<li><strong>Cold email campaigns.</strong> Multi-step sequences, A-Z testing with up to 26 variants, 8 types of conditional branch, a visual sequence builder, a spam score checker and bounce protection before send.</li>
<li><strong>LinkedIn outreach.</strong> Connection requests, messages and InMails, with a 4-week safety ramp for new accounts. No LinkedIn credentials are stored, and connecting an account happens in the app.</li>
<li><strong>Email verification.</strong> Each address comes back Valid, Risky, Invalid or Unknown, and Unknown results are free. You can verify in the app, over the REST API or as an agent tool.</li>
<li><strong>An AI agent can run it, inside the limits.</strong> Claude, ChatGPT, Cursor, Codex or any MCP agent can connect to the public MCP server and create, launch and manage campaigns. The agent never sends a message directly, and no setting, plan, support request or agent can raise a safety limit. Connecting accounts stays in the app.</li>
</ul>
<h2>WarmySender vs Instantly vs Lemwarm</h2>
<table>
<thead>
<tr>
<th></th>
<th>WarmySender</th>
<th>Instantly (as of Nov 2026)</th>
<th>Lemwarm (as of Nov 2026)</th>
</tr>
</thead>
<tbody><tr>
<td>What it is</td>
<td>Warmup + cold email + LinkedIn + verification</td>
<td>Cold email platform with warmup</td>
<td>Email warmup and deliverability tool</td>
</tr>
<tr>
<td>Email warmup</td>
<td>Peer-to-peer, on every paid plan</td>
<td>Unlimited warmup on Outbound plans</td>
<td>Core product; 20,000+ real domains in 100+ countries</td>
</tr>
<tr>
<td>Sends cold email campaigns</td>
<td>Yes</td>
<td>Yes</td>
<td>No (lemlist does; Lemwarm is included with it)</td>
</tr>
<tr>
<td>Mailboxes</td>
<td>50 / 400 / 800 by plan</td>
<td>Unlimited email accounts on Outbound plans</td>
<td>Priced per mailbox</td>
</tr>
<tr>
<td>How pricing scales</td>
<td>Email plan is account-level; LinkedIn, WhatsApp and Instagram seats are per seat</td>
<td>Outbound plan by volume, plus separate Credits plans for lead data</td>
<td>Per mailbox per month; Essential and Smart plans</td>
</tr>
<tr>
<td>Lead data</td>
<td>200M+ B2B contacts, monthly leads by plan</td>
<td>Lead Finder via Credits plans</td>
<td>Not included</td>
</tr>
<tr>
<td>LinkedIn outreach</td>
<td>Yes, per-seat add-on</td>
<td>Not covered in its plans overview</td>
<td>Not included (lemlist covers LinkedIn)</td>
</tr>
<tr>
<td>Email verification</td>
<td>Included, real-time</td>
<td>Via Credits</td>
<td>Not included</td>
</tr>
<tr>
<td>AI agent over MCP</td>
<td>Public MCP server; agent can't send directly or raise a limit</td>
<td>Not compared here</td>
<td>Not compared here</td>
</tr>
<tr>
<td>Free trial</td>
<td>7 days, no card, on Pro and Enterprise</td>
<td>Free trial: 250 uploaded contacts and 1,000 emails</td>
<td>Not stated on its homepage</td>
</tr>
</tbody></table>
<p>Sources: <a href="https://help.instantly.ai/en/articles/10273259-instantly-plans-overview">Instantly plans overview</a>, <a href="https://help.instantly.ai/en/articles/7920548-outbound-plans-comparison">Instantly outbound plans comparison</a>, <a href="https://www.lemwarm.com">lemwarm.com</a>, <a href="https://www.lemlist.com/pricing">lemlist pricing</a>, <a href="https://warmysender.com/llms.txt">WarmySender llms.txt</a>. Head-to-head pages for each tool are on WarmySender's <a href="https://warmysender.com/blog/compare">comparison hub</a>.</p>
<h2>A note on prices</h2>
<p>You'll notice I haven't put WarmySender prices in that table. That's on purpose. Prices change, and the <a href="https://warmysender.com/pricing">pricing page</a> is the only place I'd want anyone to quote them from, so check it rather than a number copied into an article, this one included. What doesn't change: annual billing is 55% off monthly, the three-year term is 70% off, and there's a 7-day free trial with no card on Pro and Enterprise.</p>
<p>For the others: Instantly's help center lists its Growth plan at $47/month for 1,000 uploaded contacts and 5,000 emails a month. Lemwarm lists Essential at $29 and Smart at $49 per mailbox per month, with lower prices on yearly billing.</p>
<h2>When to pick which</h2>
<p><strong>Pick Lemwarm if:</strong></p>
<ul>
<li>You <strong>already use lemlist</strong>. Lemwarm is included in every lemlist plan, so there's no reason to pay for warmup elsewhere.</li>
<li>You only need <strong>warmup and deliverability monitoring</strong> for a few mailboxes and are happy with the sending tool you have.</li>
</ul>
<p><strong>Pick Instantly if:</strong></p>
<ul>
<li>You run <strong>a lot of mailboxes</strong>. Instantly's Outbound plans include unlimited email accounts and unlimited warmup, while WarmySender plans cap mailboxes at 50, 400 or 800.</li>
<li>You're an agency whose work is <strong>email-only</strong> and you want a widely used, established platform.</li>
<li>You want Instantly's <strong>CRM and done-for-you mailbox setup</strong>, which it sells alongside its Outbound plans.</li>
</ul>
<p><strong>Pick WarmySender if:</strong></p>
<ul>
<li>You want <strong>warmup and sending in one subscription</strong>, instead of a warmup tool plus a sending tool.</li>
<li>You do <strong>LinkedIn outreach too</strong> and want email and LinkedIn steps in one sequence, with replies in one inbox.</li>
<li>You have a <strong>team</strong>. The email plan is account-level, so a five-person team pays the same for email as a solo founder. Only LinkedIn, WhatsApp and Instagram seats are per account.</li>
<li>You want <strong>verification and lead data</strong> in the same place you send from.</li>
<li>You want an <strong>AI agent</strong> to run campaigns in plain language, under limits it can't change.</li>
</ul>
<h2>A word on doing cold outreach responsibly</h2>
<p>No tool makes cold email safe by itself. Warmup builds sender reputation, but it won't fix a bad list or a message people didn't want. Verify addresses before you send, keep volumes steady, include a working unsubscribe, and follow the laws and provider rules that apply to you. The same goes for LinkedIn: conservative pacing lowers the risk to your account, but LinkedIn's own rules still apply, and you're responsible for following them.</p>
<h2>Track record</h2>
<p>WarmySender has also run as a lifetime deal on <a href="https://appsumo.com/products/warmysender/">AppSumo</a>, where it has done about $170K in sales. As of November 9, 2026, the listing shows a 4.8 rating from 67 reviews. Those reviews are public, so you can read what buyers liked and what they asked us to fix.</p>
<h2>FAQ</h2>
<p><strong>What is WarmySender?</strong>
An outreach platform you can hand to an AI agent. It covers cold email, peer-to-peer email warmup, LinkedIn outreach and multichannel sequences, plus real-time email verification and a 200M+ B2B lead database. Instagram and WhatsApp are per-seat add-ons.</p>
<p><strong>How does the warmup work?</strong>
Your mailboxes exchange warmup emails with other real, verified mailboxes in the network, in multi-turn threads. An adaptive engine adjusts volume based on deliverability, and mailboxes with high spam rates are throttled or removed from the pool. It works with Gmail, Outlook and SMTP.</p>
<p><strong>How is WarmySender different from a warmup-only tool like Lemwarm?</strong>
Warmup-only tools warm mailboxes and stop there, so you still need a separate sending tool. WarmySender includes cold email campaigns, A-Z testing, conditional branching and a unified inbox along with warmup.</p>
<p><strong>How is it different from Instantly?</strong>
Both send cold email and warm mailboxes. WarmySender also runs LinkedIn outreach, includes real-time verification and a B2B lead database, and prices email per account rather than by volume plus separate credit plans. Instantly includes unlimited email accounts, so it can suit very large mailbox counts better.</p>
<p><strong>Can an AI agent send emails on its own?</strong>
No. An agent can create and launch campaigns over MCP, but launching only hands the campaign to the scheduler, which paces every email and LinkedIn action within daily, weekly and hourly caps and a gradual ramp. The agent can't send directly or raise a limit.</p>
<p><strong>Is there a free trial?</strong>
Yes. Pro and Enterprise come with a 7-day free trial and no credit card, on monthly and annual billing. Current prices are on the <a href="https://warmysender.com/pricing">pricing page</a>.</p>
<h2>Try it</h2>
<p>Start the trial at <a href="https://warmysender.com">warmysender.com</a>, or browse the <a href="https://warmysender.com/blog/compare">comparison hub</a> for head-to-head pages on Instantly, Lemwarm, lemlist, Smartlead and others.</p>
<p>If you use Instantly or Lemwarm and something here is out of date, tell me and I'll update it.</p>
]]></content:encoded></item><item><title><![CDATA[A Postiz alternative you pay for once: how PostDashPro compares]]></title><description><![CDATA[PostDashPro is an AI social media scheduler and Postiz alternative where AI agents like Claude, ChatGPT, Codex and Cursor plan, write and schedule posts to…]]></description><link>https://numanhamza.hashnode.dev/a-postiz-alternative-you-pay-for-once-how-postdashpro-compares</link><guid isPermaLink="true">https://numanhamza.hashnode.dev/a-postiz-alternative-you-pay-for-once-how-postdashpro-compares</guid><category><![CDATA[social media]]></category><category><![CDATA[SaaS]]></category><category><![CDATA[AI]]></category><category><![CDATA[marketing]]></category><category><![CDATA[Productivity]]></category><dc:creator><![CDATA[Numan Hamza]]></dc:creator><pubDate>Mon, 09 Nov 2026 18:06:54 GMT</pubDate><content:encoded><![CDATA[<p><strong>PostDashPro is an AI social media scheduler and Postiz alternative where AI agents like Claude, ChatGPT, Codex and Cursor plan, write and schedule posts to 12 networks, and you review and edit everything in a visual calendar before it publishes.</strong> It's hosted, it's sold as a pay-once lifetime deal, and AI runs on your own API key, so there are no AI credits.</p>
<p>I'm Numan, and I co-founded PostDashPro with Furqan Javed in 2026. Postiz is the closest comparison for what we do: both tools let an agent schedule posts over MCP. This post sets them side by side, fairly. PostDashPro facts come from <a href="https://postdashpro.com">postdashpro.com</a> and its <a href="https://postdashpro.com/llms.txt">llms.txt</a>. Postiz facts come from <a href="https://postiz.com">postiz.com</a> and <a href="https://postiz.com/pricing">postiz.com/pricing</a> as of November 2026.</p>
<h2>Why we built it</h2>
<p>Posting one update to a dozen networks means a dozen tabs, a dozen character limits, and a dozen ways for a post to fail without anyone noticing. We wanted one composer with a live preview for each network, posts that go out on schedule, and an alert the moment one doesn't.</p>
<p>Then agents changed the job. If Claude or ChatGPT can draft a week of posts, the scheduler's job shifts. It becomes the place where you check the agent's work before anything goes live. So in PostDashPro:</p>
<ul>
<li><strong>Your agent plans and schedules; you review.</strong> Claude, ChatGPT, Codex, Cursor, OpenClaw and others connect over API and MCP on every plan. They can draft per-platform captions, add media and schedule posts, and each post shows a preview before it's scheduled.</li>
<li><strong>Every post is previewed per network.</strong> Live previews show fonts, truncation, character counters and action rows the way each app renders them. Captions are checked against each app's limits.</li>
<li><strong>AI runs on your own key.</strong> AI captions and the AI image studio use your own API key, so you pay your model provider directly and there's no monthly AI allowance to run out of.</li>
<li><strong>You hear about failures right away.</strong> Delivery alerts tell you the moment a post fails.</li>
<li><strong>Connections use each network's official sign-in.</strong> Accounts connect through OAuth or the key the network provides. No scraping, and you can disconnect any account at any time.</li>
</ul>
<h2>The 12 networks</h2>
<p>X (Twitter), LinkedIn, Instagram, Facebook, TikTok, YouTube, Threads, Pinterest, Bluesky, Telegram, Discord and Mastodon. Image, video and carousel posts work on every tier.</p>
<p>Postiz's list overlaps heavily with this one and also includes Reddit. If you need a network outside these 12, check Postiz first.</p>
<h2>PostDashPro vs Postiz: side by side</h2>
<table>
<thead>
<tr>
<th></th>
<th>PostDashPro</th>
<th>Postiz (as of Nov 2026)</th>
</tr>
</thead>
<tbody><tr>
<td>How it's sold</td>
<td>Lifetime deal, pay once, nothing renews</td>
<td>Monthly or yearly subscription</td>
</tr>
<tr>
<td>Entry price</td>
<td>$59 once (25 connected accounts)</td>
<td>Standard: $29/month (5 channels), or $23/month billed yearly</td>
</tr>
<tr>
<td>Other tiers</td>
<td>$129, $299, $499 once</td>
<td>Team $39, Pro $49, Ultimate $99 per month</td>
</tr>
<tr>
<td>Free trial</td>
<td>14 days, no card</td>
<td>7 days on every plan</td>
</tr>
<tr>
<td>Refund</td>
<td>60-day money-back guarantee on every tier</td>
<td>Up to one month of fees if the last charge was within 120 days; annual plans not refundable</td>
</tr>
<tr>
<td>Hosting</td>
<td>Fully hosted, nothing to install</td>
<td>Hosted cloud, or self-host the open-source edition (AGPL-3.0)</td>
</tr>
<tr>
<td>AI agents over MCP / API</td>
<td>Every plan</td>
<td>Every plan (MCP server, CLI, API, webhooks)</td>
</tr>
<tr>
<td>AI generation</td>
<td>AI captions and images on your own key, no credit allowance</td>
<td>Monthly AI image, AI video and video-clipping allowances by plan</td>
</tr>
<tr>
<td>Networks</td>
<td>12</td>
<td>X, LinkedIn, Instagram, Facebook, TikTok, YouTube, Threads, Pinterest, Reddit, Discord, Bluesky, Mastodon and more</td>
</tr>
<tr>
<td>Per-platform previews</td>
<td>Live, pixel-accurate previews per network</td>
<td>Not listed on its pricing page</td>
</tr>
<tr>
<td>Multi-brand / agencies</td>
<td>Brand slots (3 to 150 workspaces)</td>
<td>Customer groups</td>
</tr>
<tr>
<td>X posts</td>
<td>Capped per month by tier (10 to 50), because X charges per post</td>
<td>Unlimited posts per month listed on every plan</td>
</tr>
<tr>
<td>Extras</td>
<td>Omni-Draft, Brand Voice, Idea Board, n8n community node</td>
<td>RSS auto-post, repeated posts, Plugs, post comments, analytics, AI video clipping</td>
</tr>
</tbody></table>
<p>Sources: <a href="https://postiz.com/pricing">Postiz pricing</a>, <a href="https://postiz.com">Postiz homepage</a>, <a href="https://postdashpro.com/compare/postiz.html">PostDashPro vs Postiz</a>, <a href="https://postdashpro.com/llms.txt">PostDashPro llms.txt</a>.</p>
<h2>What each PostDashPro tier includes</h2>
<table>
<thead>
<tr>
<th>Tier</th>
<th>One payment</th>
<th>Connected accounts</th>
<th>Workspaces (brand slots)</th>
<th>AI agent connections</th>
<th>X posts/month</th>
</tr>
</thead>
<tbody><tr>
<td>Lifetime Tier 1</td>
<td>$59</td>
<td>25</td>
<td>3</td>
<td>1</td>
<td>10</td>
</tr>
<tr>
<td>Lifetime Tier 2</td>
<td>$129</td>
<td>50</td>
<td>15</td>
<td>3</td>
<td>20</td>
</tr>
<tr>
<td>Lifetime Tier 3</td>
<td>$299</td>
<td>100</td>
<td>50</td>
<td>10</td>
<td>30</td>
</tr>
<tr>
<td>Lifetime Tier 4</td>
<td>$499</td>
<td>250</td>
<td>150</td>
<td>25</td>
<td>50</td>
</tr>
</tbody></table>
<p>Every tier gets all 12 platforms, unlimited scheduled posts and drafts (X aside), AI captions and the AI image studio on your own key, and API and MCP access. Each connected profile, channel or Facebook Page counts as one account. You can upgrade later by paying only the difference.</p>
<p>Two limits worth knowing before you buy:</p>
<ul>
<li><strong>X is capped per month on every tier</strong>, because X charges per post. If X is your main channel and you post many times a day, this matters.</li>
<li><strong>Media is temporary.</strong> Uploads can be up to 200 MB per file. Files not attached to a post or saved draft are removed after 24 hours, drafts keep their media for 30 days after the last save, and posted media is removed 7 days after the post publishes. PostDashPro is a scheduler, not a media archive.</li>
</ul>
<h2>When to pick which</h2>
<p><strong>Pick Postiz if:</strong></p>
<ul>
<li>You want to <strong>self-host</strong>. Postiz publishes an open-source edition under AGPL-3.0, and PostDashPro has no self-hosted version.</li>
<li>You need <strong>Reddit</strong> or another network outside PostDashPro's 12.</li>
<li>You post to <strong>X heavily</strong>. Postiz lists unlimited posts per month, and PostDashPro caps X per tier.</li>
<li>You want <strong>RSS auto-posting, repeated evergreen posts, engagement Plugs, or AI video clipping</strong> from long YouTube videos.</li>
<li>You'd rather <strong>pay monthly</strong> and stop when you no longer need it.</li>
</ul>
<p><strong>Pick PostDashPro if:</strong></p>
<ul>
<li>You'd rather <strong>pay once</strong> than keep another monthly bill.</li>
<li>You already pay for an AI model and want to <strong>use your own key</strong> instead of a monthly credit allowance.</li>
<li>You want <strong>nothing to self-host or maintain</strong>: no server, database or queue.</li>
<li>You want to <strong>see every post rendered per network</strong> before it's scheduled, and to be told the moment one fails.</li>
<li>You run an <strong>agency or several brands</strong> and want brand slots that keep each client's accounts, voice and calendar separate.</li>
<li>You use <strong>n8n</strong>. The <code>n8n-nodes-postdashpro</code> community node can queue or draft posts to 8 networks, import media, and read the queue and analytics. Nothing publishes instantly from n8n.</li>
</ul>
<h2>Moving from Postiz</h2>
<p>Sign up, connect your social accounts with each network's official sign-in, and point your AI agent at PostDashPro. Posts already queued in Postiz stay there until they go out, so you can move over at your own pace.</p>
<h2>FAQ</h2>
<p><strong>What is PostDashPro?</strong> An AI-powered social media scheduler. AI agents plan, generate and schedule posts to 12 networks, and you review and edit everything in a visual calendar before it publishes.</p>
<p><strong>Which AI agents work with it?</strong> Claude, ChatGPT, Codex, Cursor, OpenClaw and more, over API and MCP on every plan.</p>
<p><strong>Do I pay for AI credits?</strong> No. AI runs on your own API key, so you're never billed by PostDashPro for AI usage.</p>
<p><strong>How much does it cost?</strong> Lifetime deals, paid once: $59, $129, $299 or $499. Every account starts with a 14-day free trial with no card, and every tier has a 60-day money-back guarantee.</p>
<p><strong>Do I have to self-host it?</strong> No. PostDashPro is fully hosted. Sign up, connect your accounts and start scheduling.</p>
<p><strong>Who is it for?</strong> Creators, teams and agencies who want to write once and publish everywhere. Brand slots cover agencies and creators running more than one brand.</p>
<h2>Try it</h2>
<p>Start the 14-day trial at <a href="https://postdashpro.com">postdashpro.com</a>. If you're comparing several tools, our <a href="https://postdashpro.com/compare.html">comparisons page</a> also covers Buffer, Hootsuite, Later, Publer and others, and the <a href="https://postdashpro.com/alternatives/postiz.html">Postiz alternative page</a> has the short version of this post.</p>
<p>If you use Postiz and see something here that's out of date, tell me in the comments and I'll fix it.</p>
]]></content:encoded></item><item><title><![CDATA[A Resend alternative built for apps and AI agents: how AgentiSend compares]]></title><description><![CDATA[Resend-compatible REST API, hosted MCP server, and a send budget per key]]></description><link>https://numanhamza.hashnode.dev/a-resend-alternative-built-for-apps-and-ai-agents-how-agentisend-compares</link><guid isPermaLink="true">https://numanhamza.hashnode.dev/a-resend-alternative-built-for-apps-and-ai-agents-how-agentisend-compares</guid><category><![CDATA[email]]></category><category><![CDATA[SaaS]]></category><category><![CDATA[api]]></category><category><![CDATA[Developer Tools]]></category><category><![CDATA[AI]]></category><dc:creator><![CDATA[Numan Hamza]]></dc:creator><pubDate>Mon, 09 Nov 2026 18:03:51 GMT</pubDate><content:encoded><![CDATA[<p><strong>AgentiSend is a transactional email API and Resend alternative that accepts Resend-compatible requests, puts a send budget on every API key, and gives you one call that stops all sending.</strong> It is meant for apps that send receipts, password resets, sign-in codes and alerts, and for the AI agents those apps now run.</p>
<p>I'm Numan, and I built it. This post covers why, how it lines up against Resend on price and features, and where Resend is the better choice. Resend figures below are from Resend's own pricing and docs pages as of November 2026. AgentiSend figures are from <a href="https://agentisend.com">agentisend.com</a> and its <a href="https://agentisend.com/pricing">pricing page</a>.</p>
<h2>Why I built it</h2>
<p>Resend is a good product. Plenty of developers use it, it has React Email, and its free tier is useful. I'm not writing this to knock it.</p>
<p>What pushed me was a narrower problem. More and more code that sends email isn't a person clicking a button. It's an agent or a background job calling an API in a loop. If that loop goes wrong, nobody notices until a few thousand people get the same email.</p>
<p>So I wanted an email API where the limits are part of the key itself:</p>
<ul>
<li><strong>Every new API key is created with a send budget.</strong> When the budget is spent, the API refuses with <code>agent_budget_exceeded</code>, and a key can't raise its own budget.</li>
<li><strong>One request stops everything.</strong> <code>POST /limits/kill-all</code> pauses every key on the account. To stop one key, use <code>POST /limits/keys/{id}/kill</code>. Both write an audit row.</li>
<li><strong>Repeated sends that look like a loop are held.</strong> They wait in an approval queue until a person decides.</li>
<li><strong>Every refusal tells you what to do next.</strong> The error carries a <code>code</code>, a <code>message</code>, a <code>fix</code> naming the call that repairs it, a <code>docs_url</code>, and whether it's <code>retryable</code>.</li>
<li><strong>You can test without sending.</strong> <code>POST /emails/preflight</code> runs every check a real send runs (domain, sandbox, suppression, budget, loop, content) and sends nothing.</li>
</ul>
<p>The same operations and limits are available over REST and through a hosted MCP server at <code>https://api.agentisend.com/mcp</code>, so Claude, Cursor or any MCP client can send under the same rules as your code.</p>
<h2>Switching from Resend: the request shape is the same</h2>
<p>This was a hard requirement for me. If you already use Resend, moving shouldn't mean rewriting your call sites.</p>
<p><code>POST /emails</code>, <code>POST /emails/batch</code>, <code>GET/PATCH /emails/{id}</code> and <code>POST /emails/{id}/cancel</code> accept Resend's request bodies, camelCase names included. You can point the official Resend Node SDK at AgentiSend by changing the base URL and the key:</p>
<pre><code class="language-js">import { Resend } from 'resend';

const resend = new Resend(process.env.AGENTISEND_API_KEY, {
  baseUrl: 'https://api.agentisend.com',
});

await resend.emails.send({
  from: 'Acme &lt;receipts@acme.com&gt;',
  to: 'customer@example.com',
  subject: 'Your receipt',
  html: '&lt;p&gt;Thanks for your order.&lt;/p&gt;',
});
</code></pre>
<p>For Python, set <code>RESEND_API_URL=https://api.agentisend.com</code>. There's also a compatibility client in the Node package (<code>agentisend/compat-resend</code>) if you'd rather change only the import line.</p>
<p>Two things differ on purpose, and the <a href="https://agentisend.com/docs/guides/migrating-from-resend">migration guide</a> says so up front. First, keys have enforced budgets, so if you port a key that sends heavily, set its budget to your real volume before you cut over. Otherwise the first thing the migration does is stop. Second, domains have a region setting (<code>us</code> or <code>eu</code>) that defaults to <code>us</code>. Today all mail is still sent from US nodes, and <code>eu</code> domains move to EU sending when it ships.</p>
<h2>AgentiSend vs Resend: side by side</h2>
<table>
<thead>
<tr>
<th></th>
<th>AgentiSend</th>
<th>Resend (as of Nov 2026)</th>
</tr>
</thead>
<tbody><tr>
<td>Free plan</td>
<td>5,000 emails/month, 170/day, 3 domains, no card, no expiry</td>
<td>3,000 emails/month, 100/day, 3 domains, free forever</td>
</tr>
<tr>
<td>First paid plan</td>
<td>Starter: $9/month for 30,000 emails</td>
<td>Pro: $20/month for 50,000 emails (no plan between Free and Pro)</td>
</tr>
<tr>
<td>50,000 emails/month</td>
<td>Pro: $16/month</td>
<td>Pro: $20/month</td>
</tr>
<tr>
<td>100,000 emails/month</td>
<td>Pro: $28/month</td>
<td>Pro: $35/month, or Scale: $90/month</td>
</tr>
<tr>
<td>Yearly billing</td>
<td>40% off every paid plan</td>
<td>Monthly only, no annual discount</td>
</tr>
<tr>
<td>Pay-once option</td>
<td>Lifetime plans from $49 for 250,000 emails that don't expire</td>
<td>Not offered</td>
</tr>
<tr>
<td>Spending limit per key</td>
<td>Send budget on every key, enforced</td>
<td>API key permissions; no per-key send budget listed</td>
</tr>
<tr>
<td>Idempotency keys</td>
<td>Every mutating endpoint, kept 7 days</td>
<td>Sends, batches and some inbox actions, kept 24 hours</td>
</tr>
<tr>
<td>Message history</td>
<td>7 / 30 / 90 / 180 days by plan</td>
<td>30 days on Free, Pro and Scale</td>
</tr>
<tr>
<td>Default rate limit</td>
<td>600 requests per key per 60 seconds</td>
<td>10 requests per second per team</td>
</tr>
<tr>
<td>MCP server</td>
<td>Hosted, same operations and limits as REST</td>
<td>Remote and self-hosted MCP server</td>
</tr>
<tr>
<td>SMTP relay</td>
<td>Not documented; REST and MCP only</td>
<td>Yes, every plan</td>
</tr>
<tr>
<td>React Email</td>
<td>Not a replacement for it</td>
<td>Yes</td>
</tr>
<tr>
<td>Inbound email</td>
<td>Per-domain switch with one MX record</td>
<td>Every plan, plus 3 inboxes</td>
</tr>
<tr>
<td>Security reports</td>
<td>No SOC 2 report yet</td>
<td>SOC 2 Type II on every plan</td>
</tr>
<tr>
<td>DKIM key</td>
<td>2048-bit RSA</td>
<td>1024-bit (Resend explains why it considers this enough)</td>
</tr>
<tr>
<td>Dedicated IP</td>
<td>On request on Scale</td>
<td>$30/month add-on on Scale, for senders over 3,000 emails/day</td>
</tr>
<tr>
<td>Uptime commitment</td>
<td>99.9% monthly on Scale</td>
<td>99.99% SLA on Enterprise</td>
</tr>
</tbody></table>
<p>Sources: <a href="https://resend.com/pricing">Resend pricing</a>, <a href="https://resend.com/docs/dashboard/emails/idempotency-keys">Resend idempotency keys</a>, <a href="https://resend.com/docs/api-reference/rate-limit">Resend rate limits</a>, <a href="https://resend.com/docs/knowledge-base/do-i-need-2048-dkim">Resend on 2048-bit DKIM</a>, <a href="https://resend.com/llms.txt">Resend llms.txt</a>. AgentiSend: <a href="https://agentisend.com/pricing">pricing</a>, <a href="https://agentisend.com/docs/guides/migrating-from-resend">migration guide</a>, <a href="https://agentisend.com/llms.txt">llms.txt</a>. The full, sourced comparison lives at <a href="https://agentisend.com/vs/resend">agentisend.com/vs/resend</a>.</p>
<p>A note on the price rows: at each volume Resend sells, AgentiSend's monthly price is 20% lower ($16 vs $20 at 50,000, $28 vs $35 at 100,000). The bigger difference is in billing terms. Resend is monthly only. AgentiSend has a 40% yearly discount and lifetime plans: $49 for 250,000 emails, $99 for 600,000, $199 for 1.5M and $349 for 3M. Each is sent at up to a monthly maximum, and the emails never expire.</p>
<h2>How the limits behave when you hit them</h2>
<p>I care about this more than the headline price, because it's what happens at 3 a.m.</p>
<p>On AgentiSend, sending pauses when you reach your plan's emails. Nothing extra is charged unless an owner turns overage on in billing settings and sets a ceiling. Overage is $0.75 per 1,000 on Starter, $0.72 on Pro and $0.60 on Scale, and sending stops at the ceiling you set. The free plan never charges overage.</p>
<p>On Resend, pay-as-you-go lets a paid plan keep sending past its quota at $0.90 per 1,000 on Pro. If you haven't turned it on, Resend notifies you and offers an upgrade.</p>
<p>Neither approach is wrong. If a hard stop would hurt you more than an unexpected bill, Resend's default may suit you better. If an unexpected bill is the bigger risk, especially with an agent holding the key, a hard stop is the safer default.</p>
<h2>When to pick which</h2>
<p><strong>Pick Resend if:</strong></p>
<ul>
<li>You build emails with <strong>React Email</strong> and want that workflow end to end.</li>
<li>You need <strong>SMTP</strong> (a CMS, a legacy app, or a tool that only speaks SMTP).</li>
<li>Your buyers ask for a <strong>SOC 2 Type II</strong> report. AgentiSend doesn't have one yet, and its <a href="https://agentisend.com/security">security page</a> says so.</li>
<li>You want <strong>inboxes for agents</strong> with threads, or a <strong>dedicated IP</strong> at a published $30/month.</li>
<li>You want the better-known API with a bigger ecosystem around it. That counts for something, and AgentiSend is the newer product.</li>
</ul>
<p><strong>Pick AgentiSend if:</strong></p>
<ul>
<li>An AI agent or automated job holds the API key, and you want a <strong>send budget on that key</strong>, a <strong>kill switch</strong>, and <strong>loop detection</strong> without building them yourself.</li>
<li><strong>Price</strong> matters: a larger free plan, 20% less at every volume Resend sells, yearly billing, or a pay-once lifetime plan.</li>
<li>You want <strong>longer message history</strong> on paid plans (90 days on Pro, 180 on Scale).</li>
<li>You want refusals your code, or your agent, can act on, because every error names its fix.</li>
</ul>
<p>One more thing to be clear about: AgentiSend is for email people expect to get. Cold email isn't a supported use, and the API refuses sends that look like it. Newsletters and announcements to people who opted in are fine, with topics and one-click unsubscribe headers built in.</p>
<h2>FAQ</h2>
<p><strong>Is AgentiSend a drop-in replacement for Resend?</strong> For sending, mostly yes. The core email endpoints accept Resend's request bodies, and the official Resend SDKs work when you change the base URL. Errors come back with Resend's top-level <code>name</code>, <code>message</code> and <code>statusCode</code>, plus AgentiSend's own <code>error.code</code> and <code>error.fix</code>. It isn't a replacement for React Email, and it doesn't document SMTP submission.</p>
<p><strong>Is there a free plan?</strong> Yes. Every account starts on Free: 5,000 emails a month, up to 170 a day, 3 sending domains, no card and no expiry. Test sends don't count toward it.</p>
<p><strong>How much does AgentiSend cost after the free plan?</strong> Starter is $9/month (30,000 emails), Pro is $16/month (50,000) or $28/month (100,000), and Scale starts at $72/month (100,000). Yearly billing takes 40% off. Lifetime plans run from $49 once for 250,000 emails.</p>
<p><strong>How do I stop an AI agent from sending too much email?</strong> Give the agent its own API key. Every new key is created with a budget. The key can't raise its own budget, and <code>POST /limits/kill-all</code> stops every key on the account. Sends that look like a loop wait for a person to approve them.</p>
<p><strong>Can I keep data in the EU?</strong> You can set a domain's region to <code>eu</code>. As of now, all mail is still sent from US nodes, and <code>eu</code> domains move to EU sending automatically when it ships.</p>
<p><strong>Does AgentiSend have an MCP server?</strong> Yes, hosted at <code>https://api.agentisend.com/mcp</code> (streamable HTTP, bearer auth). It has the same operations and limits as the REST API, so an agent can't do anything over MCP that the key's budget wouldn't allow over REST.</p>
<h2>Try it</h2>
<p>The free plan doesn't need a card, so the fastest way to judge it is to send something. Start at <a href="https://agentisend.com">agentisend.com</a>, read the <a href="https://agentisend.com/docs">docs</a>, or go straight to the <a href="https://agentisend.com/vs/resend">AgentiSend vs Resend comparison</a> if you want every row with its source.</p>
<p>If you're on Resend and something in this comparison is wrong or out of date, tell me in the comments and I'll fix it. Pricing pages change, and I'd rather be corrected than quoted wrong.</p>
]]></content:encoded></item></channel></rss>