PitchButler
P1
Replies rot in the inbox.
Sending is automated; reading is not. Interested people wait days for an answer and the deal cools before anyone notices. The money is lost after the reply, not before it.
YOUR AUTOMATED SDR - COLD EMAIL TO BOOKED MEETING
Leads. Served on a hot plate.
PitchButler is an automated SDR: it finds the right person, emails them from your own Gmail, follows up, reads every reply - and books the meeting. Your team has one job: show up and close.
Limited intake - 30 new teams a week. Waitlist open now.
One-to-one email only - no blast machinery to misuse
Your mailbox, your domain, your sent folder
Pauses itself at the first sign of trouble
Watch it work: reply in read & scored handed to a human.
An awkward thing to admit
We don’t want your team using this platform.
Every hour inside a sales tool is an hour not selling. So PitchButler runs the outreach itself - and only taps you when there’s a meeting to take.
Success isn’t your team logging in. It’s them never needing to.
The uncomfortable metric we optimise for
We understand your pain
Cold outreach is broken in three predictable ways.
Every team that comes to us has hit the same three walls. None of them is fixed by sending more email.
P2
Blast tools burn your domain.
Volume-first tools push hundreds of sends a day from a cold mailbox. Gmail notices, deliverability collapses, and every future email pays for it, including the ones that matter.
P3
The pipeline lives in five tabs.
The sequencer, the inbox, the spreadsheet, the CRM and someone's memory all disagree. Nobody can say which leads are hot right now, so follow-up depends on luck.
House rules
Four rules. Zero exceptions.
These aren’t settings you can fumble - they’re welded into the engine.
01
Nothing ever sends twice.
Every email becomes a ticket with a unique receipt before it can fire. Duplicates aren’t filtered out - they’re structurally impossible.
02
“No” means forever.
An opt-out is permanent, has no undo, and is checked before every single
send. Literally: do_not_contact = false.
03
It never guesses.
A reply it can’t place with confidence goes to a human review queue - not to a coin flip. Ambiguity is escalated, never improvised.
04
Your name comes first.
Bounce or spam-complaint spikes trip a kill switch that pauses all sending - before your domain pays for it, not after.
How it works
Interested
Not interested
Unclear
Follow one lead through the machine.
The easiest way to understand PitchButler is to watch it work a single lead - call her Jane, at Acme. You give it a name. It does the patient, repetitive work a salesperson hates, and comes back when there’s something worth your time.
1
You add a company
Acme Corp - Jane Lindqvist is enough. No email address required, no enrichment subscription, no spreadsheet gymnastics.
2
It finds Jane
The discovery pipeline searches, reads Acme’s website, cross-checks data providers, and verifies the address - before a single word is sent to it.
3
It does the homework
Acme gets read before it gets written to. The first line of the intro comes from research on the company - not from a mail-merge template.
4
A considered intro
One email, to one person, sent from your own Gmail, inside your working hours, minutes apart from the last one. The way you would send it.
5
Patient follow-ups
No reply? It waits the interval you chose, then nudges politely. It never crowds, and it stops the moment Jane writes back.
6
Jane replies
The reply desk files her message to the right conversation - matched on hard evidence, never guessed - reads it, and scores how interested she really is.
Step 7 - every reply leaves through one of three doors
You’re tapped on the shoulder.
An instant alert with the thread, the intent score, and what Jane cares about. Automation stops; the human conversation starts.
Never contacted again.
One “no thanks” places Jane on a permanent do-not-contact list, checked before every future send. Forever, with no undo.
A human decides.
Anything ambiguous, legal, or commercial lands in a review queue with the AI’s reasoning attached. It asks; it does not improvise.
The reply desk
It reads replies the way you would.
Sending is the easy half - most tools stop there. PitchButler’s real work begins when the answer comes back: every reply is matched to the right conversation, read by Claude, Anthropic’s frontier model, and routed to one of sixteen specialist playbooks.
Filed on evidence
A five-step matcher binds each inbound to its thread using message headers, thread IDs, and quoted text - hard signals first, in strict order. When the evidence is thin, the reply goes to review instead of being filed on a hunch.
Scored, not skimmed
Promising replies get a second read: how genuine is the interest, what are the objections, is it time for a human? Cross the threshold and your phone buzzes. Fall short and a courteous follow-up is scheduled instead.
Sensitive means human
Legal questions, procurement gates, complaints, anything the model flags about its own output - these are never AI-answered. They go straight to you, with context.
JL
Today, 09:41
Jane Lindqvist
jane.lindqvist@acme.se
Re: Acme × PitchButler - quick intro
Hi Anna - this actually looks relevant for us. We’re reviewing suppliers this quarter. Could you do a short call on Thursday afternoon? /Jane
Read & routed - 09:42
The CRM
Your pipeline lives here, not in five tabs.
PitchButler isn’t a plugin that fires emails into the void - it’s the system of record. Every lead, every touch, every reply, and everything the AI has learned about a person sits in one place, kept current without you doing the data entry.
A
HOT
Jane Lindqvist
Acme Corp · acme.se
-
Mar 12, 09:14Introduction sentStep 1 of 4 · via anna@yourco.se
-
Mar 19, 10:02Follow-up sentStep 2 of 4 · gentle nudge
-
Mar 21, 09:41Reply received“This actually looks relevant for us…”
-
Mar 21, 09:42Read & qualifiedInterested · intent 87/100 · handoff recommended
-
Mar 21, 09:42Handed to youAutomation stopped · you replied at 10:05
What the AI remembers
- Reviewing fleet-tracking suppliers this quarter - decision before summer.
- Prefers concrete pricing early; allergic to vague “let’s hop on a call.”
- Final sign-off goes through the operations director (Karl).
Memory is built from the conversation itself and follows the lead - so the human who takes over starts informed, not cold.
A timeline you never have to type
Every send, reply, classification, and handoff is logged the moment it happens. The record is the system’s own ledger - not someone’s memory of last Tuesday.
Memory that travels with the lead
The AI keeps note of what each person cares about, what they objected to, and who actually decides. When a lead goes hot, you inherit the context - not just an email address.
Queues, not chaos
Ambiguous matches, drafts awaiting approval, and stuck leads each have their own queue with a clear next action. The pipeline tells you where you’re needed; nothing depends on you noticing.
The engine room
Built to know when to stop.
Deliverability is existential in cold email - one burned domain costs more than any meeting is worth. So the engine treats your sender reputation as the first thing to protect, and restraint as a feature, not a setting.
S-01
The warm-up ramp
A new mailbox starts at five emails a day and earns its way up over weeks - like easing into the gym. Gmail never sees a cold mailbox behave suspiciously.
S-02
Human pacing
Sends are spaced minutes apart, inside the working hours and weekdays you choose. There is no setting for “blast 500 at midnight” - on purpose.
S-03
The kill switch
Bounce and spam-complaint rates are watched continuously. A spike pauses all outbound automatically - and stays paused until a human reviews it.
S-04
Domain watch
SPF, DKIM, DMARC and blocklist status are checked daily on every sending domain. A listing pauses the account before the damage compounds.
S-05
The permanent no
Every send path filters do_not_contact = false. An opt-out is
enforced at the database - not in a settings page someone forgot.
S-06
The ledger
Every action is written down as a ticket before it happens. If a worker crashes mid-task, the ticket retries safely - nothing fires twice, nothing is silently lost.
The warm-up ramp - emails per day, per mailbox
default schedule · day 1 15
d1d2d3d4d5d6d7d8d9d10d11d12d13d14d15
The actual default schedule. Small numbers are the point - reputation is earned slowly.
The intelligence
The smartest model first. A human before a bad guess.
The thinking is done by Claude, Anthropic’s frontier models. Every reply gets the best reader available - and when nothing good is available, PitchButler would rather wait than improvise. Prefer sign-off on every word? Approval mode has the AI draft while you approve each send.
First
The careful reader
Claude Opus
Reads and scores every reply, catches nuance, drafts with judgment. The default for everything customer-facing.
If unavailable
The reliable second
Claude Sonnet
Steps in when the primary is busy. Fast, dependable, and still held to the same guardrails.
If still unsure
You
The human safety net
If no model is confident, nothing is sent. The task waits, or lands in your queue with the reasoning attached.
“A weak first impression is worse than a slightly delayed one.”
The design principle behind every fallback
The night shift
14 handled · 2 for you
A night manager that does the rounds.
Every few minutes, the Steward sweeps the queues an operator would otherwise hand-triage - unconfirmed bounces, stalled leads, identity puzzles. It resolves the obvious and leaves the rest in your morning digest, with its reasoning attached.
Handles the obvious
An out-of-office gets parked until the return date. A confirmed bounce gets quarantined. A follow-up that drifted out of its window gets rescheduled. None of this should ever have needed you.
Escalates with its work shown
Anything genuinely ambiguous arrives in the digest as a question, not a mystery: what it found, what it considered, and why it stopped short of acting. You decide in seconds, not minutes.
Bounded on purpose
The Steward works from a fixed, audited list of allowed actions per situation. It’s a diligent night manager - not an unsupervised intern with admin access.
While you were out
Tuesday 06:00 · overnight sweep
- Parked 3 out-of-office replies until their stated return dates.
- Confirmed the bounce at nordvik.se and quarantined the address.
- Rescheduled 9 follow-ups into today’s send window.
- Matched a reply from a new address back to Meridian Freight - same person, new domain.
- !Needs you: a reply from helena@cedar.co fits two open threads - pick one and I’ll do the rest.
- !Needs you: a draft to Stenmark Bygg is waiting for your approval.
Why teams switch
Typical cold email tools vs PitchButler.
Typical cold email tools
- Automate the sending, stop at the reply
- Reward volume: more contacts, more blasts, more risk
- Send from rented domains and shared pools
- Guess when a reply is ambiguous
- Leave the pipeline scattered across tabs and spreadsheets
PitchButler
- Reads and scores every reply, hands hot leads to a human in minutes
- One-to-one email at human pace; there is no blast button to misuse
- Sends from your own Gmail, on your own domain, into your own sent folder
- Escalates ambiguity to a review queue; it never improvises
- Is the system of record: every lead, touch and reply in one place
Pricing
Pay for work done, not for seats.
PitchButler runs on credits: you top up, and the engine spends them on the research, writing and reply-reading it actually does. No per-seat licences, no annual lock-in, and a hard budget cap so a quiet month costs almost nothing.
No-spin answers
The things you should ask any tool like this.
What does my team actually do in the platform?
As little as possible - by design. The engine does the finding, the homework, the writing, the sending, the reading and the filing. Your team shows up for the hot leads and the occasional judgment call, and that’s it. If you find yourself spending your day inside PitchButler, write to us - something is wrong.
Is this just another mass-mailer?
No - structurally, not just philosophically. PitchButler sends one-to-one email, like a person typing to one other person. There is no bulk-list machinery, no “upload 50,000 contacts” button, and no way to blast. That single choice shapes everything else about how it’s built.
Does it send from my domain?
From your own Gmail mailbox, through Google’s official API. Your name, your address, your sent folder - replies land in your inbox like any other conversation. No shared sending pools, no rented domains.
Will this burn my domain?
It’s engineered so the answer is no: a slow warm-up ramp, human pacing inside your chosen hours, daily SPF/DKIM/DMARC and blocklist checks, and a kill switch that pauses everything at the first bounce or complaint spike. The system’s bias is to send less, not more.
Why is there a waitlist?
Because warm-up is real and hands-on. A brand-new mailbox can’t safely send at volume on day one - it has to earn its reputation slowly, and we watch that ramp for every account. So we deliberately take on about 30 new teams a week, no more, rather than onboard everyone at once and give each one a rushed, unsupervised start that puts their domain at risk. The waitlist isn’t a growth tactic - it’s the same “send less, not more” instinct that runs through the whole product. Join it and we take you in turn.
What happens when someone opts out?
They are never contacted again. The opt-out is permanent, has no undo, and
is enforced at the database - every send path filters
do_not_contact = false. This is the backbone of the whole system,
not a checkbox.
Which AI does the thinking?
Claude, by Anthropic - the Opus model first, Sonnet as the reliable second. If neither is confident, nothing is sent: the task waits or goes to a human. There is no “cheap model for the customers” tier, and no canned replies pretending to be AI.
Can I approve every email before it sends?
Yes. Approval mode has the AI research, draft, and queue - while you approve, edit, or reject each send. Many teams start there and loosen the reins once they trust the output.
What does it do when it isn’t sure?
It asks. An ambiguous reply match goes to a review queue where you bind it to the right conversation in one click - and the pipeline replays from there as if the match had been clean. Guessing is how automated email gets you in trouble, so it never does.
How many emails a day will it send?
Fewer than you’d think - and that’s the point. A fresh mailbox starts at around five a day and ramps to about fifty over weeks. Tools that promise thousands a day are describing the speed at which they’ll burn your domain.
More questions? Read the full FAQ or write to hello@pitchbutler.io.
Who’s behind it
There is one person on the other side of this.
Olle Nilsson
Founder · PitchButler
Fourth company
Founding since 19
Built in Sweden
Hi - I’m Olle, and I write the code you’ve just been reading about.
PitchButler is my fourth company. The first was a foodtech startup I started at nineteen, backed by Swedish operators including the founder of RGNT Motorcycles. What stayed with me across all of them wasn’t any funding round. It was the outreach: years of writing to people who had never heard of us, and the slow discovery that the only emails that ever worked were the ones where I’d done real homework first.
Every tool I tried pushed me the other way - more contacts, more volume, more risk to a domain I couldn’t afford to burn. So I built the thing I actually wanted instead: something that does the reading before it writes, sends one email to one person from my own mailbox, and taps me on the shoulder only when someone is genuinely interested.
That’s also why the waitlist is real and the pace is slow. I’d rather warm up thirty mailboxes properly than three hundred badly. If you join, I’m the one reading it - and if something breaks, I’m the one fixing it.
Olle
By invitation · 30 teams a week
Put your outreach
Put your outreach
in good hands.
We open 30 new mailboxes a week - no more. Each one gets a real, supervised warm-up before it sends in earnest, and that hands-on care is exactly what keeps your domain safe. So we run a waitlist and take teams in the order they join.
It’s the same instinct as the rest of the product: do it properly, not quickly. Join the list and we’ll tell you where you stand and when your warm-up begins.
You’re on the list.
We’ll be in touch before your warm-up slot opens.
already a customer? sign in · or write to hello@pitchbutler.io