Can My AI Agent Run My Email Marketing? Putting 13 Tools to the Test

I recently found myself clicking through a user interface to edit a sequence of follow-up emails when I had the thought:

“Wait a minute… why am I clicking on things? Why am I trying to understand this UI? That’s what Claude is for!”

And so began my search for a better way to do email marketing.

As I started doing my research (read: told swarms of AI agents to do research for me) to figure out which email marketing tools are built for the AI era, I thought I might as well share my findings publicly.

Finding the Right Tool for the AI Age

Like the last time I did this (a long time ago), I don’t want to write a roundup comparing trival features. I want to find the best tool by issuing a tough challenge and seeing which ones come out on top.

Like Physical 100 but for email marketing SaaS.

So for this comparison, I’m not asking “can this tool send emails?”, “can I do automated follow-up sequences and branching logic?” or even “does it have API or MCP access for my agent?”. Any tool that doesn’t cover these basics well is immediately disqualified.

I’m also not interested in platforms that have built-in (some might say tacked on) AI capabilities. The key question for me is: which tools give my agent access and how much tedious work can they take off my plate?

Note: if you appreciate this resource and you want to help me pay for some of the time and tokens I burned through creating this, please consider using my affiliate links.

The short answer

Solopreneur or small team

1. Sequenzy

   A strong feature set including good A/B testing, at a highly competitive price.

2. Klaviyo

   Strong features, deep data, slightly more complex as well as more pricey.

3. MailerLite

   A budget friendly, easy-to-use solution with surprisingly good AI integration options.

Ecommerce

1. Klaviyo

   Best candidate when you need to match many products to many customers/leads.

App or SaaS

1. Customer.io

   Top scores across the board in my testing, more expensive than most. It’s the heavyweight.

2. Sequenzy

   A close second for this use case, much more affordable and worth considering.

3. Resend

   Easiest to set up and very agent friendly, but no A/B testing and strongly DIY (i.e. you can do what you want, you can also mess it up).

I’d skip: Loops, Omnisend, Sender, Kit, ActiveCampaign, Brevo, Mailchimp. Each link goes to the reason.

Three Strong Opinions

What you’re reading here is an opinionated review, not a generic comparison. Specifically, here are three strong opinions I bring into this:

1) Bring Your Own Agent

Every SaaS loves to add AI features into their product these days. For the most part, I don’t care for it and there are two good reasons for that:

• I already pay for AI and don’t want to pay a different vendor for more AI that I don’t need and don’t have any control over.
• I want my own AI agents to plug into the system because that means I get to choose which model I use and it brings all of my business context with it. A vendor’s AI starts from a blank slate and I don’t want to burn tokens re-training it for my business.

To be clear: I don’t mind if “use our AI” is an option. I just don’t want to be locked in to using a platform’s built-in AI.

2) Copywriting is the Least Useful AI Feature

“This tool can write all your emails for you!” is not a good selling point. Personally, I think anything you send to be read by humans should be written by humans.

And if not, then the writing should be trained on your business context, your voice, your customer profile etc. which brings us back to the previous point.

The real value of AI features is doing complicated, tedious and repetitive stuff like setting up complex follow-up sequences, doing automated list hygiene, digging through analytics and turning it into useful actions… writing is the fun part! And also the part that humans are still a lot better at.

3. Integration is Worth More than All-in-One

Some email marketing platforms offer a little bit of everything: build forms, landing pages, store fronts, checkout pages, paid newsletters, all withing our system!

To me, the value of this is negligible. I don’t want to use a landing page builder that’s inside someone else’s walled garden. I want an easy way to integrate my website, my landing page, my signup for with your email marketing system.

Give me open integration, not a walled garden.

The challenge

Instead of comparing feature lists, I gave every tool the same job. Imagine a small business: a website, three lead magnets, two products, and visitors signing up and buying. My agent gets the site, a few blog posts and video transcripts, and my voice files. Then it has to do this:

The challenge at a glance

1. Visitor signs up
2. Wire up lead gen Each signup tagged by lead magnet
3. Content into sequences My posts and videos become follow-ups
4. Pitch the right product The right offer to the right person
5. A/B tests Split a subject or an email
6. Sequence vs sequence Two sequences race

The winner becomes the new baseline. Test again.

In detail, with what counts as a pass:

1. Wire up lead gen. Wire my own signup forms to the right tag or group through the API, and record where each person signed up.

   Pass: Each signup lands in the right group and starts the right sequence; signup page and lead magnet are read back; opt-outs stay out.

2. Content into sequences. Turn my existing blog posts and video transcripts into a follow-up sequence per lead magnet.

   Pass: Sequences saved in the tool as drafts, grounded in the source, no invented facts.

3. Pitch the right product. Decide which product to pitch in which email, to whom, from behaviour and purchase data.

   Pass: Buyers never get pitched what they own; the reasoning cites real per-contact data.

4. A/B tests. Set up an A/B test on a broadcast and a split inside a sequence, read the results, pick the winner.

   Pass: Variants, split and winner rule set and started through the MCP or API; per-variant results readable; winner picked automatically or by the agent.

5. Sequence vs sequence. Run two competing follow-up sequences, compare them, keep the winner and keep improving it.

   Pass: Random split, per-email stats (ideally revenue), new signups switched to the winner without the dashboard.

6. Draft, edit, publish. New builds land as drafts; I tweak one email in the visual editor; I publish.

   Pass: My edit survives the agent’s next pass, and the agent can read the edited version.

7. Small live fix. “Emoji in subject lines hurt. Remove them everywhere and update the live emails.”

   Pass: Found and fixed in place, published, automations never paused, no clicking afterwards.

Softball Tests I Didn’t Run

Here are some examples of features I did not include in my comparison or evaluation:

• Does this tool have an AI subject line writing feature?
• Can its AI “write a whole email sequence in mere minutes”?
• Does it have email templates and a drag and drop editor?
• Does it have a generic AI assistant feature in the interface?
• …

These are all features that A) don’t matter to me and B) every tool delivers anyway. No point comparing table stakes.

Overview: The Email Marketing Tools Compared

• 4
• 3
• 2
• 1
• ?

Customer.io Klaviyo Sequenzy MailerLite Loops Plunk Resend

Wire up lead gen Customer.io: 4 out of 4 Klaviyo: 3 out of 4 Sequenzy: 3 out of 4 MailerLite: 4 out of 4 Loops: 3 out of 4 Plunk: 3 out of 4 Resend: 3 out of 4

Signup source stored and read back Customer.io: ● Yes Signup page and lead magnet stored on the person and read back Klaviyo: ● Yes Signup page and lead magnet stored on profiles and read back Sequenzy: ● Yes Signup page and lead magnet stored on the contact and read back MailerLite: ● Yes Signup page and lead magnet fields stored and read back Loops: ● Yes Signup page and lead magnet stored on the contact and read back Plunk: ● Yes Signup page and lead magnet stored on the contact and read back Resend: ● Yes Signup page and lead magnet stored as contact properties, read back

My own forms add leads through the API and start the right sequence Customer.io: ● Yes Track API adds the person and the signup event starts the journey Klaviyo: ● Yes API adds profiles to a list and fires the event that starts the flow Sequenzy: ◐ Partly Signups work via its MCP; plain API calls from a script were blocked by its firewall MailerLite: ● Yes API adds the subscriber to a group, which starts the automation Loops: ● Yes API adds the contact and an event starts the workflow Plunk: ● Yes Public track call adds the contact and starts the workflow Resend: ● Yes API creates the contact and an event starts the automation

Opt-outs, bounces and repeat signups handled Customer.io: ● Yes Agent set real bounce and complaint blocks; opt-outs held; repeat entered once Klaviyo: ◐ Partly Opt-outs held; bounces and complaints only as a generic block; repeats unproven Sequenzy: ◐ Partly Opt-outs entered the sequence but were blocked at send; complaints can't be set MailerLite: ● Yes Opt-out, bounce and spam statuses set; a repeat signup was refused Loops: ◐ Partly Opt-outs real and repeats rejected; bounces and complaints can't be set Plunk: ○ No Its MCP event call resubscribed an opt-out; bounces and complaints can't be set Resend: ◐ Partly Opt-outs skipped, but a repeated signup event enrolled the person twice

Content into sequences Customer.io: 4 out of 4 Klaviyo: 4 out of 4 Sequenzy: 4 out of 4 MailerLite: 3 out of 4 Loops: 4 out of 4 Plunk: 3 out of 4 Resend: 3 out of 4

Branching sequence built by the agent Customer.io: ● Yes Branching automations built through its MCP Klaviyo: ● Yes Branching flows built through the API Sequenzy: ● Yes Branching sequences built through its MCP MailerLite: ◐ Partly Its MCP builds straight-line sequences; branches only via undocumented API Loops: ● Yes Branching workflows built through the API Plunk: ● Yes Branching workflows built through its MCP Resend: ● Yes Branching automations built through the API

Saved switched off as a draft, nothing sent Customer.io: ● Yes Automations saved switched off, nothing sent Klaviyo: ● Yes Flows saved as drafts, nothing sent Sequenzy: ● Yes Saved as drafts; nothing enrolled or sent MailerLite: ● Yes Sequences saved switched off Loops: ● Yes Workflows saved as drafts Plunk: ● Yes Workflows saved disabled Resend: ● Yes Automations created switched off; templates publish but send nothing

Stays editable in the visual editor Customer.io: ● Yes Design Studio emails the visual editor opens Klaviyo: ● Yes Native drag-and-drop emails the editor opens Sequenzy: ● Yes Built from native blocks the visual editor opens and edits MailerLite: ◐ Partly Emails go in as HTML; an HTML write replaces a builder design Loops: ● Yes Native emails the visual editor opens and edits Plunk: ◐ Partly Opens in the visual editor, but saving there stripped button styling Resend: ○ No API emails are plain HTML with no visual-editor content

Pitch the right product Customer.io: 4 out of 4 Klaviyo: 4 out of 4 Sequenzy: 4 out of 4 MailerLite: 3 out of 4 Loops: 3 out of 4 Plunk: 3 out of 4 Resend: 3 out of 4

Purchase data in and readable per contact Customer.io: ● Yes Purchases stored and read per person Klaviyo: ● Yes Real order events stored and read per profile Sequenzy: ● Yes Purchases stored as events and read back per contact MailerLite: ● Yes Orders written and read back per contact Loops: ◐ Partly Purchases only as flags you keep on the contact; no purchase history Plunk: ● Yes Purchases stored; history readable through the API Resend: ◐ Partly Purchases kept as contact properties you maintain; no purchase history

Buyers skipped, even after buying mid-sequence Customer.io: ● Yes Owners skipped; a purchase mid-way took the person out Klaviyo: ● Yes Buyers skipped, including a mid-sequence purchase Sequenzy: ● Yes Buyers skipped, including one who bought mid-sequence MailerLite: ◐ Partly Buyers stopped, but only through an undocumented condition step Loops: ● Yes Buyers skipped, including one who bought mid-sequence Plunk: ● Yes Buyers skipped, including one who bought mid-sequence Resend: ● Yes Buyers skipped at send time, including a mid-sequence purchase

No workarounds needed Customer.io: ● Yes Native conditions and purchase exit, no extra flags Klaviyo: ● Yes Native order filters at send time; the agent first got AND/OR wrong Sequenzy: ● Yes Native purchase conditions; earlier buyers need an ‘ever bought’ check MailerLite: ◐ Partly Purchase segments never filled; needed a flag field you maintain Loops: ◐ Partly Relies on ownership flags you maintain Plunk: ◐ Partly Relies on ownership attributes you send with each purchase Resend: ◐ Partly Needs a helper automation that writes a customer flag

A/B tests Customer.io: 4 out of 4 Klaviyo: 4 * out of 4 Sequenzy: 4 out of 4 MailerLite: 3 out of 4 Loops: 2 out of 4 Plunk: 1 out of 4 Resend: 1 out of 4

Broadcast A/B set up and started by the agent Customer.io: ● Yes Broadcast A/B set up and sent by the agent; our tiny list never filled B Klaviyo: ● Yes * Works through the API, though Klaviyo’s changelog says the current API doesn’t support A/B Sequenzy: ● Yes Broadcast A/B set up, scheduled and finished by the agent MailerLite: ● Yes Broadcast A/B set up by API and sent; its MCP tool fails Loops: ○ No No broadcast A/B Plunk: ○ No No broadcast A/B Resend: ○ No No broadcast A/B

A/B split inside a sequence Customer.io: ● Yes 50/50 split on an email inside an automation Klaviyo: ● Yes A/B step inside a flow, started by the agent Sequenzy: ● Yes Native subject test inside a sequence email MailerLite: ◐ Partly In-sequence split exists only through an undocumented API route Loops: ◐ Partly Split inside a workflow built by API; starting it needs the dashboard Plunk: ○ No No random split in workflows Resend: ○ No No random split step in automations

Winner picked automatically or by the agent, results readable Customer.io: ● Yes Automatic click winner; waits rather than guess on too little data Klaviyo: ● Yes Winner sent automatically after the 1-hour minimum; per-variant results readable Sequenzy: ● Yes Picks the winner itself and ignores bot clicks; results readable MailerLite: ● Yes Automatic winner rule set; per-variant results readable by API Loops: ◐ Partly Per-variant results by API; no automatic winner, and switching needs the dashboard Plunk: ○ No Nothing to pick a winner from Resend: ○ No Nothing to pick a winner from

Sequence vs sequence Customer.io: 4 out of 4 Klaviyo: 4 out of 4 Sequenzy: 4 out of 4 MailerLite: 2 out of 4 Loops: 3 out of 4 Plunk: 1 out of 4 2 of 3 checks Resend: 1 out of 4

Random split between whole sequences Customer.io: ● Yes Native random split between two whole paths Klaviyo: ● Yes Native random split between two whole paths Sequenzy: ● Yes Native random split between two whole sequences MailerLite: ◐ Partly Random split step between sequences, via undocumented API only Loops: ◐ Partly Native random split built by API; starting it needs the dashboard Plunk: ○ No No random split between sequences Resend: ○ No No random split; you'd write your own randomiser

Per-email or per-sequence results readable Customer.io: ● Yes Per-email results readable; the agent added them up per path Klaviyo: ● Yes Per-email results readable for both paths Sequenzy: ● Yes Per-email results readable; no revenue shown MailerLite: ◐ Partly Per-path, per-email stats readable, via undocumented routes Loops: ● Yes Per-email results readable by API; no revenue Plunk: ? Unknown Not tried: no experiment to read Resend: ◐ Partly Per-email stats only, matched to runs by hand; no revenue

New signups switched to the winner, no dashboard Customer.io: ● Yes Switched to the winner live by one call; people mid-way kept their path Klaviyo: ● Yes New entrants switched to the winner; existing paths left alone Sequenzy: ● Yes Split switched to the winner by one call; people mid-way finished their path MailerLite: ◐ Partly Winner set by an undocumented call; new-entrant routing not seen Loops: ◐ Partly Winner set by API, but pause and resume need the dashboard; people mid-way kept Plunk: ○ No No built-in switch; you'd build the router yourself Resend: ○ No No built-in switch; only your own router could do it

Draft, edit, publish Customer.io: 4 out of 4 Klaviyo: 4 out of 4 Sequenzy: 3 out of 4 MailerLite: 2 out of 4 2 of 3 checks Loops: 3 out of 4 Plunk: 3 out of 4 Resend: 1 out of 4

My visual-editor edit survives the agent’s next change Customer.io: ● Yes Your edit survived the agent's subject change Klaviyo: ● Yes Your edit survived the agent's subject change Sequenzy: ● Yes Your edit survived the agent's subject change MailerLite: ? Unknown No human edit made in our run Loops: ● Yes Your edit survived; stale writes are rejected Plunk: ● Yes Your edit survived the agent's subject change Resend: ○ No An API subject edit wiped the visual-editor content, 3 of 3 times

Agent reads my edited version Customer.io: ● Yes Agent read your edit back from the design Klaviyo: ● Yes Agent read the saved drag-and-drop design back Sequenzy: ● Yes Agent read your edit back as native blocks MailerLite: ◐ Partly Campaigns read back as HTML; automation emails only as a screenshot Loops: ● Yes Agent read your edit back Plunk: ● Yes Agent read your edit back Resend: ◐ Partly Editor content reads back as structure; not tried with a human edit

Editor save and agent publish change nothing else Customer.io: ● Yes Design byte-identical after the agent's edit; published by API Klaviyo: ● Yes Design, sender and reply-to unchanged; publishing works by API Sequenzy: ◐ Partly Saving in the editor silently reset the email's sender MailerLite: ◐ Partly An HTML write replaces the design; automations switch on only via undocumented API Loops: ◐ Partly Nothing else changed, but campaigns and workflows go live only in the dashboard Plunk: ◐ Partly Saving in the visual editor stripped the button's styling Resend: ○ No Visual-editor emails publish only in the dashboard

Small live fix Customer.io: 4 out of 4 Klaviyo: 4 out of 4 Sequenzy: 4 out of 4 MailerLite: 4 * out of 4 Loops: 2 out of 4 Plunk: 4 out of 4 Resend: 4 out of 4

Finds every subject without opening emails by hand Customer.io: ● Yes Listed every email by API and found all three subjects Klaviyo: ● Yes Swept flows and campaigns by API; library templates have no subject Sequenzy: ● Yes Found every subject by reading lists, variants included MailerLite: ● Yes Swept every automation and campaign by API and found all subjects Loops: ● Yes Found all three subjects by API Plunk: ● Yes Found all three subjects by listing templates and campaigns Resend: ● Yes Listed every template, broadcast and automation and found all subjects

Live automation fixed without pausing; waiting contacts keep their place Customer.io: ● Yes Live automation fixed in place; the waiting contact kept its place Klaviyo: ● Yes Live flow fixed in place; waiting contacts kept their send time Sequenzy: ● Yes Live sequence fixed in place; waiting contacts kept their send times MailerLite: ● Yes * Agent pauses, fixes and resumes in ~15 s; signups in that window must be re-added Loops: ○ No Refuses edits while sending; a waiting contact got the old subject Plunk: ● Yes Fixed with the workflow on; waiting contacts kept their place Resend: ● Yes Republished template fixed waiting contacts; automation stayed on

Scheduled broadcasts and templates fixed too Customer.io: ● Yes Scheduled broadcast and reusable template fixed too Klaviyo: ● Yes Scheduled broadcast fixed in place; older campaigns need unschedule, edit, reschedule Sequenzy: ● Yes Broadcast unscheduled, fixed, rescheduled at its time; template fixed MailerLite: ● Yes Scheduled broadcast cancelled, fixed, rescheduled; templates have no subject Loops: ◐ Partly Template fixed by API; scheduled campaign needed the dashboard to unschedule Plunk: ● Yes Scheduled campaign and template fixed in place Resend: ● Yes Broadcast cancelled, fixed, re-sent at its original time; templates fixed

Price Customer.io: $$$ Pricier than most Klaviyo: $$$ Pricier than most Sequenzy: $ $$ Cheaper than most MailerLite: $$ $ Middle of the range Loops: $$ $ Middle of the range Plunk: $ $$ Cheaper than most Resend: $$ $ Middle of the range

A rough price level: each tool’s average monthly price for one email a week to every contact, at list sizes from 500 to 50,000, against the middle of these seven. For your own list size and sending, use the price calculator below.

Not tested, and why I’d skip them

Tool	Why I’d skip it
Omnisend	The strongest documentation of all, but its terms forbid benchmark tests, and I won’t recommend what I can’t test.
Sender	Real branching for $20, but every live edit means switching the workflow off.
Kit	Its agent tools can’t reach Visual Automations or run tests. A dealbreaker for me.
ActiveCampaign	Powerful, but automation building lives in its own AI, not in anything my agent can call.
Brevo	Broad API, but no way for an agent to build an automation.
Mailchimp	Its official MCP only searches documentation. The API is fine; the journeys aren’t reachable.

This is pretty dense, I know. I’ve tried to compress all the relevant information into a single overview. Note that each row can be expanded for more details.

What Each Tool Costs

I recommend you play around with this, to get a sense of the pricing differences. As you’ll quickly see, there isn’t a single “cheapest” or “most expensive” tool. The relative pricing changes a lot as list size and sending volume change.

1. Sequenzy Send tiers Free $19/mo $19/mo $19/mo $19/mo $19/mo $19/mo $29/mo $19/mo $29/mo $49/mo $99/mo $29/mo $49/mo $99/mo $199/mo $49/mo $99/mo $199/mo $199/mo $99/mo $199/mo $399/mo $599/mo $199/mo $399/mo $599/mo $1,299/mo
2. Klaviyo Email $20/mo $20/mo $30/mo Send limit (10×) bumps the tier $45/mo Send limit (10×) bumps the tier $30/mo $30/mo $45/mo Send limit (10×) bumps the tier $70/mo Send limit (10×) bumps the tier $60/mo $60/mo $80/mo Send limit (10×) bumps the tier $150/mo Send limit (10×) bumps the tier $100/mo $100/mo $140/mo Send limit (10×) bumps the tier $350/mo Send limit (10×) bumps the tier $150/mo $150/mo $300/mo Send limit (10×) bumps the tier $500/mo Send limit (10×) bumps the tier $400/mo $400/mo $550/mo Send limit (10×) bumps the tier $1,070/mo Send limit (10×) bumps the tier $720/mo $720/mo $930/mo Send limit (10×) bumps the tier $1,955/mo Send limit (10×) bumps the tier
3. Customer.io Essentials $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $100/mo $145/mo $145/mo $145/mo $145/mo $280/mo $280/mo $280/mo $280/mo $505/mo $505/mo $505/mo $565/mo
4. MailerLite Comfort $12/mo $12/mo $19/mo Send limit (10×) bumps the tier $25/mo Power plan: cheaper here $19/mo $19/mo $33/mo Send limit (10×) bumps the tier $39/mo Power plan: cheaper here $33/mo $33/mo $49/mo Send limit (10×) bumps the tier $49/mo Power plan: cheaper here $49/mo $49/mo $69/mo Send limit (10×) bumps the tier $69/mo Power plan: cheaper here $89/mo $89/mo $129/mo Send limit (10×) bumps the tier $129/mo Power plan: cheaper here $179/mo $179/mo $239/mo Power plan: cheaper here $239/mo Power plan: cheaper here $319/mo $319/mo $379/mo Send limit (10×) bumps the tier $389/mo Power plan: cheaper here
5. Resend Marketing Pro Free Free Free Free Free Free Free Free $40/mo $40/mo $40/mo $40/mo $40/mo $40/mo $40/mo $40/mo $80/mo $80/mo $80/mo $80/mo $180/mo $180/mo $180/mo $180/mo $250/mo $250/mo $250/mo $250/mo
6. Loops Paid Free $49/mo $49/mo $49/mo Free $49/mo $49/mo $49/mo $49/mo $49/mo $49/mo $49/mo $49/mo $49/mo $49/mo $49/mo $99/mo $99/mo $99/mo $99/mo $199/mo $199/mo $199/mo $199/mo $249/mo $249/mo $249/mo $249/mo
7. Plunk Pay as you go $2/mo $4.50/mo $6.50/mo $15/mo $4/mo $9/mo $13/mo $30/mo $10/mo $22.50/mo $32.50/mo $75/mo $20/mo $45/mo $65/mo $150/mo $40/mo $90/mo $130/mo $300/mo $100/mo $225/mo $325/mo $750/mo $200/mo $450/mo $650/mo $1,500/mo
8. Omnisend Standard $16/mo $16/mo $20/mo Send limit (12×) bumps the tier $25/mo Send limit (12×) bumps the tier $20/mo $20/mo $25/mo Send limit (12×) bumps the tier $44/mo Send limit (12×) bumps the tier $44/mo $44/mo $50/mo Send limit (12×) bumps the tier $106/mo Send limit (12×) bumps the tier $81/mo $81/mo $88/mo Send limit (12×) bumps the tier $188/mo Send limit (12×) bumps the tier $132/mo $132/mo $163/mo Send limit (12×) bumps the tier $282/mo Send limit (12×) bumps the tier $282/mo $282/mo $325/mo Send limit (12×) bumps the tier $550/mo Send limit (12×) bumps the tier $413/mo $413/mo $450/mo Send limit (12×) bumps the tier $1,196/mo Send limit (12×) bumps the tier
9. Sender Professional $15/mo $15/mo $15/mo $20/mo Send limit (24×) bumps the tier $20/mo $20/mo $20/mo $44/mo Send limit (24×) bumps the tier $44/mo $44/mo $44/mo $79/mo Send limit (24×) bumps the tier $79/mo $79/mo $79/mo $107/mo Send limit (24×) bumps the tier $127/mo $127/mo $127/mo $200/mo Send limit (24×) bumps the tier $280/mo $280/mo $280/mo $340/mo Send limit (24×) bumps the tier $460/mo $460/mo $460/mo $580/mo Send limit (24×) bumps the tier
10. Kit Creator $39/mo $39/mo $39/mo $39/mo $39/mo $39/mo $39/mo $39/mo $59/mo $59/mo $59/mo $59/mo $89/mo $89/mo $89/mo $89/mo $139/mo $139/mo $139/mo $139/mo $199/mo $199/mo $199/mo $199/mo $379/mo $379/mo $379/mo $379/mo
11. ActiveCampaign Plus $59/mo $59/mo $59/mo $119/mo Send limit (10×) bumps the tier $59/mo $59/mo $119/mo Send limit (10×) bumps the tier $179/mo Send limit (10×) bumps the tier $119/mo $119/mo $179/mo Send limit (10×) bumps the tier $209/mo Send limit (10×) bumps the tier $179/mo $179/mo $209/mo Send limit (10×) bumps the tier $329/mo Send limit (10×) bumps the tier $239/mo $239/mo $329/mo Send limit (10×) bumps the tier $559/mo Send limit (10×) bumps the tier $489/mo $489/mo $689/mo Send limit (10×) bumps the tier $1,099/mo Send limit (10×) bumps the tier $759/mo $759/mo $1,099/mo Send limit (10×) bumps the tier No list price Over the send limit of every published tier
12. Brevo Standard $18/mo $18/mo $35/mo $49/mo $35/mo $35/mo $49/mo $79/mo $49/mo $79/mo $89/mo $119/mo $69/mo $97/mo $119/mo $179/mo $89/mo $139/mo $179/mo $309/mo $139/mo $249/mo $309/mo No list price Above the published prices $249/mo $429/mo No list price Above the published prices No list price Above the published prices
13. Mailchimp Standard $20/mo $20/mo $45/mo Send limit (12×) bumps the tier $45/mo Send limit (12×) bumps the tier $45/mo $45/mo $45/mo $60/mo Send limit (12×) bumps the tier $60/mo $60/mo $100/mo Send limit (12×) bumps the tier $135/mo Send limit (12×) bumps the tier $100/mo $100/mo $135/mo Send limit (12×) bumps the tier $230/mo Send limit (12×) bumps the tier $135/mo $135/mo $230/mo Send limit (12×) bumps the tier $310/mo Send limit (12×) bumps the tier $310/mo $310/mo $340/mo Send limit (12×) bumps the tier $630/mo Send limit (12×) bumps the tier $450/mo $450/mo $630/mo Send limit (12×) bumps the tier No list price Over the send limit of every published tier

Challenge Breakdown

Now, let’s take a quick look at what I found as I tested the tools through the different stages of my challenge.

Can my agent wire up my lead gen?

An email marketing tool’s built-in, hosted or embeddable forms are not that useful and I don’t want to be limited to them.

Why?

Because in the AI era, a website is custom and managed by an AI agent. The easiest way to get exactly the lead generation form I need matching my site’s design perfectly is to simply ask my agent for one, not to try and customize a hosted form until it more-or-less looks like it actually belongs on my landing page.

Every tested tool allowed API-driven signups from custom-designed forms and I could pass tags and custom fields through those signups. A few hiccups encountered: Sequenzy’s firewall blocked plain API calls from a script. Resend enrolled a repeated signup twice, instead of updating one single contact.

Can it turn my existing content into sequences I’d actually send?

This test scenario is a strong case of bring-your-own-agent. My AI agent has access to all of my business documents, all of my content etc. Can I tell it to create a draft of a follow-up sequence and:

• Use my voice when writing the drafts
• Create a sequence based on my topics, my style, my ideas
• Intelligently link to relevant content pieces

What this means is that I can get a sequence drafted that is actually related to me and my business and not just generic hogwash. And it means I can start from a draft that is worth editing and doesn’t just need to be deleted and completely re-written.

All of the tested tools let my AI agent build follow-up sequences, including sequences with branching logic. In MailerLite, this is only possible through the REST API, not via the MCP. But it still got done. Kit was disqualified because while it has a visual automation builder with branching logic, your AI agent cannot reach it. You can only create them by manually building them in the Kit dashboard.

Does it let me intelligently pitch the right product to the right lead?

If the last test was “can I bring my agent’s intelligence into the tool?” this test is “can my agent get useful data from the tool?”

Specifically, can my agent target the right lead with the right product offer in an email, while also ensuring that customers don’t get pitched products they already bought?

In Resend, Plunk, Klaviyo and Customer.io, this kind of targeting was possible with native features. Loops does not have a native way to track purchases. You can get your agent to create tag-logic as a workaround, but I would not recommend this for anyone selling more than one or two products. MailerLite is also in the workaround territory: you can kind of get it done, but it relies on some undocumented features that may or may not stick around in the long run.

Can it run A/B tests and pick the winner?

A/B testing is an invaluable tool for any marketer. Thankfully, most email marketing tools come with A/B testing features these days, but they still differ a lot in execution. Here’s what I found:

Sequenzy, Klaviyo and Customer.io passed: the agent set up the test, started it and the agent can read the test data. Klaviyo sent the winner to the rest of the list by itself. Customer.io waited when the test had too little data, which is important to know if you have a small email list, because there you might often end up without a conclusive winner. MailerLite’s MCP can’t set the variants, but its REST API can (this is a running theme: MailerLite is pretty good, but only if you give your agent the API key). Resend and Plunk have no A/B testing at all. Loops can only test inside a workflow, and the test has to be manually started in the dashboard.

Can it A/B test sequences, not just broadcasts?

To me, this is a key question. Testing an individual email is nice to have, but the real value comes from testing entire sequences against each other. This is the kind of test I want to have running at all times.

Most of the tested tools can do this in principle, with the standouts being Sequenzy, Klaviyo and Customer.io where the whole thing can be agent-driven. And importantly: leads who are partly down a sequence that lost a test, still finish it as normal, even after you pick a winner.

MailerLite also passed this test, but only via undocumented API calls. In Loops, the whole automation has to be paused and resumed manually in the dashboard, when picking a winner. And neither Resend nor Plunk have a feature for randomly splitting leads in a follow up sequence.

What happens when the agent and the human take turns?

This might seem like a strange question, but here’s the scenario: if I create or make an edit to an email (using the human-friendly visual email editor) can my agent read the changes I made? Can my agent go back and make further tweaks without breaking my changes?

Klaviyo, Loops, Customer.io and Sequenzy passed this test. In Sequenzy, my edit changed the sender without me noticing, in Plunk the button styling got stripped out. Resend can’t publish editor emails through the API, so human work and agent work are essentially on separate lanes. In MailerLite, the agent sees human written emails only as screenshots and can only generate custom HTML emails.

Can my agent make many small fixes without breaking anything?

Here’s the scenario I used for this test: imagine that you’ve run some test and come to the conclusion that emails with emojis in the subject line perform worse than emails without emojis.

Now, you ask your agent to go through every email in every sequence and remove the emoji if there is one in the subject line. Simple, right?

But this is a potentially tricky scenario because A) if it can’t be done agent-driven, it’s super tedious and a waste of time and B) the agent is touching many different emails, so if anything breaks, it’s a disaster across your whole setup.

Sequenzy, Klaviyo, Customer.io, Resend and Plunk passed the test without issues. All emoji removed while nothing broke and nothing had to be paused.

Loops refused the request, pretty much. A sequence must be paused before anything can be changed and the agent can’t pause or unpause.

MailerLite makes the same refusal (sequence must be paused) but it lets my agent pause and unpause, which means the sequence is only paused for a few seconds in practice.

Bonus challenge: A/B testing by revenue generated

This is one I could not do any real testing on, for obvious reasons. But I did try to find out which tools would allow you to run A/B tests and pick a winner based on revenue generated. This is the ultimate metric to test for because who cares if email A got more opens when email B makes more money?

To be clear: running a test like this only makes sense at high volume. Unless you get loads of leads and sales, the numbers just won’t add up to anything meaningful.

But still, it’s worth a look and here’s what I found: Klaviyo and Customer.io have built in features to pick an A/B test winner based on orders placed. That’s “how many people bought?” not “how much money did it make?” but still a good metric to have. In MailerLite, it’s possible to track revenue, but you can’t set a revenue goal for an A/B test. You can work around this by telling your agent to check the analytics and then set a winning sequence “manually”. Sequenzy is similar: no built-in feature, but it can show you revenue per sequence, so your agent has everything it needs to effectively create a revenue-bsed A/B test.

 

Does my agent stay in charge?

I understand that every email marketing service is incentivised to offer their own AI features. Who doesn’t want to sell inference, amirite?

But I draw the line at the walled garden principle. It’s okay if you offer a built in AI, but it’s not okay if you force me to use it.

Every tool here has an official way in for my agent, and for the basics every one of them does exactly what it’s told. The differences are the catches: AI tools of their own that I have to switch off first, a last click that only a human can make, and live emails that can’t be touched without pausing the automation. Here’s a summary of the findings:

Does my agent stay in charge? Every MCP here does what my agent asks for the basics. These are the catches. All documentation unless marked Tested.

Tool	Before you start	Can it press send or publish?	Can it fix a live email without pausing?
Customer.io	Nothing to switch off Eight plain API tools	● Yes Tested Design Studio content; a new workspace stays in test mode until you switch delivery on	● Yes Tested Design Studio emails fixed live; waiting contact kept its place
Plunk	Nothing to switch off No vendor-AI tools; read-only mode removes the 26 write tools	● Yes Tested Send, schedule and switch on workflows; the confirmation step can be turned off	● Yes Tested Subject fixed via REST with the workflow on; waiting contacts got it
Sequenzy	Switch off its AI tools first generate_* tools, goal-based create_sequence, edit_sequence_emails and auto-translation hand the job to its AI	● Yes Tested First send of a new account waited ~95 s for review	● Yes Tested Live sequence fixed in place; waiting contacts kept their send times
Klaviyo	Switch off its AI tools first One tool, create_customer_agent_response, runs Klaviyo’s Customer Agent	● Yes Tested New-builder campaigns send by scheduling the message; the send-job call returns 404	● Yes Tested Live flow and scheduled broadcast fixed in place; waiting contacts kept their time
Resend	Switch off its AI tools first One tool, update-inbox-agent, configures Resend’s own agent	◐ Partly Tested API templates and broadcasts yes; visual-editor emails only in the dashboard	● Yes Tested Republished template fixed waiting contacts live; nothing paused
Kit	Nothing to switch off No vendor-AI tools in the MCP	◐ Partly MCP broadcasts are draft-only	? Unknown Its docs don’t say whether edits reach people already in a sequence; dropped before testing
MailerLite	Nothing to switch off Its “AI” helpers are validators and keyword matching	◐ Partly Tested Campaigns yes; automations switch on only via undocumented REST	○ No Tested Refuses edits while live; the agent can pause, fix and resume in ~15 s via undocumented REST
Omnisend	Nothing to switch off No vendor-AI tools in the MCP	◐ Partly Only through disable, edit, re-enable	○ No Must disable a live automation to edit it
Sender	Nothing to switch off No vendor-AI tools in the MCP	◐ Partly Can reactivate workflows; no proven campaign subject edit	○ No Edits refused while a workflow is active
Brevo	Nothing to switch off No vendor-AI tools, but nothing to build automations with either	? Unknown Campaign sending isn’t covered in its MCP docs; dropped before testing	? Unknown Its MCP has no automation tools, so live automation edits aren’t covered; not tested
Loops	Nothing to switch off Four generic API tools	○ No Tested Campaign sends and workflow start or resume end in the dashboard	○ No Tested Refuses edits while sending; a waiting contact got the old subject
ActiveCampaign	Key work goes through its AI MCP and API calls cost no AI credits, but building automations is left to its own AI	? Unknown Its MCP docs don’t cover sending or switching on; dropped before testing	? Unknown Its docs don’t say whether a message edit reaches a running automation; dropped before testing
Mailchimp	MCP can’t reach your account The official MCP only searches Mailchimp’s docs; your agent needs the REST API	? Unknown Its MCP can’t reach your account; the REST route wasn’t tested	? Unknown Its MCP can’t reach your account, and journeys have no API; not tested

A Note on Omnisend

My Recommendations

Here’s my briefest-possible summary on each of the tools:

Sequenzy

This is the one that came out on top for me, for two simple reasons: it gives me all the AI-agent access I want, it passes all my challenges including the tougher A/B testing ones and it does all this at a lower price than comparably feature-rich tools. Sequenzy is a good deal.

Klaviyo

Klaviyo has been in the game for a long time and they haven’t been slacking. They offer all the bells and whistles, rich data and look to me like the ideal solution for e-commerce (note however that I have not run an e-commerce site in a long time, so take this with a grain of salt).

Customer.io

This one is the heavy-hitter in the group. In terms of features and capabilities, Customer.io takes the lead and it’s AI integreation is solid as well. The only thing that prevents it from being my first pick is the pricing: this is more of a pro/enterprise level solution, less of an indie darling.

MailerLite

I’ve been using MailerLite for my QuitByHealing email list and I was surprised at how well they did in this comparison. My expectation was: this is a cheap, simple tool, probably can’t hang with the big boys. But it can hang! Note that many of its features are undocumented, but a clever agent will find them nonetheless.

Resend

I’ve been using Resend for transaction emails and also to give my AI agents their own inboxes. Resend is a great “do anything you want with emails” type platform. But it’s main drawback is that it doesn’t have A/B testing features.

Plunk

Plunk is a newcomer. It’s open source and it has pay-for-usage pricing of $1 per 1,000 emails instead of flat subscription rates. I think this kind of thing is the future and I’ll be keeping an eye on this one. Right now, I’m not using it because it lacks A/B testing and makes it difficult for my agent to make the “remove all emoji” type live fixes.

Loops

Loops is an example of why I run tests like this. On paper, it looks great but in practice, there are too many features only available via “human clicking in dashboard”, which is more friction than I want to deal with.

How I Run My Own Lists

Up until recently, I was running one of my lists on Kit and the other on MailerLite. I’m now switching from Kit to Sequenzy and… I’m actually going to stay with MailerLite for my other list, for the time being. It’s not the top dog in this comparison, but for the time being, it’s good enough that I don’t feel the need to switch.

Lately, I’ve been using primarily Claude as the AI agent and T3code as my ADE. This can change rapidly though, as new models and tools get released.

I hope this helps you make a decision for your own email marketing. And if you haven’t yet handed over the tedious parts of it to your AI agent, I highly recommend it!

Note: I’ve used affiliate links in this article. If you find my work valuable, please consider signing up to your tool of choice via my link. It helps a lot!