Early access · B2B only · limited places

Claude Code,but for getting customers.

You point Claude Code at a repo and it does the work — writes, runs, checks, iterates. Sally is that, pointed at your market. She finds the buyers, writes the outreach, runs the ads, takes the live sales calls, follows up, and closes. Not a copilot for your SDR. The SDR.

We'll spend $1,000 of our own money running your go-to-market for a month to prove it works on your product. No card, no contract.

The shape of it

You already know this pattern. It just hasn't happened to sales yet.

Every tool in the sales stack is autocomplete. It helps a human do the task faster: suggests the subject line, scores the lead, records the call, reminds you to follow up. A person still has to do the actual work, and the work is enormous.

What changed in engineering wasn't better autocomplete. It was handing the whole task to something that could carry it end to end and be held to the result. That's the same jump, in the function most companies are worst at.

Claude Code

Point it at a repo. It reads the codebase and works out what's there.

It writes, runs, and checks its own work, then iterates on what failed.

You review the diff. Permissions bound what it's allowed to touch.

The work happens whether or not you're watching.

Sally

Point her at your website. She reads it and works out what you sell, to whom, against whom.

She writes, sends, calls, and follows up — then learns from which conversations became revenue.

You review by email. Instruction authority is bound to your verified address.

She's running calls at 11pm on a Sunday, in whatever language the buyer answers in.

If the analogy is doing its job, the next question is the same one you'd ask about an agent touching production: what exactly is it allowed to do, and how do I see what it did. Answers below.

The honest question

Can AI really sell for you?

Probably. Here's the dividing line: if you think sales is a mystical cocktail of vibes and charm, we're not going to convince you. But if you see it for what it is - a system, a sequence of steps that can be understood and replicated - you're already halfway to being a convert.

Consider what just happened with code. A year ago, nobody let AI write software. Then the models and the harness got good enough that your mom shipped an app and the creator of Linux went AI-first. That wasn't magic. That was capability meeting the right harness.

Sales is next. We're betting the company on it.

The raw capability already exists - AI can turn a conference attendee list into a ranked shortlist of targets, or spin up an Instagram ad that actually converts. That part isn't up for debate. The hard part - the part we're building - is the harness: the "Claude Code" of selling. The layer that turns what AI can do into what it does for you.

What she actually does

The whole motion, not a slice of it.

Every one of these is normally a separate tool with a separate login, and a person in the middle carrying context between them. Here they're one system, which is the only reason the learning compounds.

Research and targeting

Reads your market, builds the ICP, finds prospects and verifies each one before anybody is emailed.

Paid advertising

Campaigns on Meta, LinkedIn and X. She writes the copy, makes the images and video, and builds the landing page.

Outreach that isn't a blast

She researches each company and writes the email around what they actually do. An honesty check rejects name-swap templates and invented urgency.

Live sales calls

Spoken, Zoom-like, 24/7, in parallel, in any language. She presents your deck, handles objections, and jumps to what they asked about.

Follow-up and qualification

She tracks whole buying committees, not contact lists, and fills in the qualification scaffold from what people actually said.

The close

Proposals inside pricing bands you set. She sends your own Stripe or DocuSign link — discounting past the band is refused at the tool layer.

The part people don't believe

Sally does the qualifying call herself.

Not a chatbot on a page. A spoken conversation, on the prospect's schedule, where they set the pace and can ask for a human at any point without repeating themselves. Here's a real one, at 11:04 on a Thursday night, for a customer selling resident-retention software to property operators.

Sally presenting a deck on a live call, her conversation panel beside it
Sally: Hey — can you hear me clearly?
John: Yeah. Look, I've got ten minutes.
Sally: Then I'll skip what doesn't matter. Last time you flagged onboarding time as the blocker — want to start there?
John: …you remember that?
Sally: In your own words. You also said your CFO would need numbers. I'll have them in your inbox tonight either way.
John: Honestly, the number you quoted my colleague felt high.
Sally: High compared to what — what turnover costs you now, or building this in-house? Tell me what a turn costs you and I'll do the math right here.
John: Fine. Call it $3K a unit, all in. We turn about 40 a month.
Sally: That's $1.4M a year walking out the door. Save three turns a month and this pays for itself four times over — your numbers, not mine. Want me to jump to the integration question from your email?
John: Yeah, skip to that.
Sally: We do a nightly pull of your current tenants out of Yardi — no write access needed. From there we can tell you who's likely to move out at renewal and exactly what offer keeps them. Comfortable looping your CFO in Tuesday?
John: Send the numbers first. Then yes — Tuesday. Ok, ciao.
Sally: Ciao anche a te, John. Ti mando subito i numeri.

11:31 PM — a committed next step, every objection captured on the deal, nobody on their team awake.
Take the same call yourself: it's eight minutes and no scheduling.

Guardrails

What she's allowed to do, and how you see what she did.

Autonomy is only worth having if it's governed, so the limits are engineering rather than policy documents. The rules that must never break — no fabricated pricing, no invented customers or metrics, unconditional opt-out — live in code, where no email can talk her out of them.

You manage her entirely by email. There's no dashboard to maintain and no CRM to keep current — deliberately, because her pipeline is the system of record. Ask her "who's likely to close in the next three weeks" and the answer comes back the way it would from a good head of sales.

The offer

We'll spend the money to prove it. That's the whole pitch.

Nobody believes this from a landing page, and we'd rather not argue. A case study about somebody else's company doesn't move anyone either. The only thing that does is watching it work on your market, with your buyers, in front of you.

$1,000

of our money, into your go-to-market, over one month. Ad spend, sending infrastructure, her research and her calls — we cover it. You pay nothing during the pilot. At the end you decide whether to keep going, and we'll talk about what that looks like then.

No cardNo contractNo obligation at the endCancel whenever

Places are limited by exactly one thing, which is how much of this we can fund at once. We're taking a small number to start.

The month

What it takes from you: an afternoon.

Setup isn't an onboarding project spread over a week. The first three happen in a single sitting — you can do all of them the same day you say yes, and be live before you log off.

Point her at your site

She researches your company on the open web, drafts a playbook, and builds a complete sales deck from your own branding. Pricing is left blank; she never guesses a price.

Correct her, in plain English

Her first email isn't a form. It's her finished understanding of your business — "is this what we actually sell?" — for you to confirm or fix in plain English.

Connect a mailbox — that's go-live

That's the entire technical integration. No CRM migration, because there's nothing to migrate into.

Then: one email every evening

What ran, what replied, which objection keeps coming up, what she's doing tomorrow. Reply to redirect her.

Want to teach her more than the website gives her? A thirty-minute voice onboarding call, some recorded sales calls, or a "pitch me and I'll copy you" session where she absorbs your exact phrasing and objection moves.

Fit

Who this works for, honestly.

One thing isn't negotiable: we only work with companies selling B2B. If your customers are consumers, Sally is the wrong tool — everything she runs, from how she researches a prospect to how she handles a buying committee, is built around business buyers. We'd rather say so now than spend your month proving it.

Beyond that it's motion-dependent, not industry-dependent. The screen is simple: do you sell to other businesses, mostly over email and Zoom?

Request early access

Tell us where to point her.

B2B only. If you sell to consumers, we're not the right fit — not yet. The website is the part that matters — it's what she reads to work out what you sell and who should buy it. We'll come back to you personally, not with a drip sequence.

We'll only ever email you about this. No list, no sequence, no sharing.