Dollars in.Deals out.Humans optional.

This is Sally at 11:04 on a Thursday night.

Not our pipeline — a customer's. They sell resident retention to property operators. Their prospect finally got a quiet minute, opened his email, and tapped a link. No calendar dance. He was in a live, spoken sales call in under a second.

Thirty years of sales software, and the actual work is still done by hand.

For the last thirty years, companies have been told software would transform sales and marketing. In one sense it did. There is now a tool for almost every individual task:

They don't form a system. They form a pile.

Generating revenue is often the hardest, most expensive thing a company does. Tens of thousands of dollars a month on tools designed to find customers — and the actual work of finding customers is still overwhelmingly manual.

Each tool performs one narrow function. Each contains one fragment of the truth. Each requires someone to decide what happens next. The advertising platform can tell you who clicked an ad, but not what those people later said in a sales conversation. The outbound platform can tell you who replied, but not why one message resonated and another failed. The CRM can tell you an opportunity closed, but not which idea, objection, campaign or conversation closed it.

The salesperson knows what prospects are saying. The marketer knows which campaigns get attention. The founder knows why customers ultimately buy. The knowledge lives in different people and different systems.

And every specialist in the pile optimizes for what their piece can see. The ad agency — a monthly retainer plus ten percent of ad spend, the standard arrangement for twenty-plus years — optimizes cost-per-click and click-through rate. The email operation, in-house or hired, optimizes sends, open rates, positive-reply rates. All real metrics. None of them the one that moves the needle. Nobody is optimizing for revenue, because nobody can see all the way to it.

Humans are the glue holding the entire operation together.

• Someone has to decide who to target
• Someone has to turn positioning into campaigns
• Someone has to write the ads
• Someone has to build the lists
• Someone has to launch the outreach
• Someone has to monitor performance
• Someone has to respond to interest
• Someone has to qualify the prospect
• Someone has to follow up
• Someone has to update the CRM
• Someone has to explain the results
• Someone has to apply the learning to the next campaign

This is what companies call a sales and marketing operation. But it isn't an operation. It's a chain of disconnected tools, held together by people — expensive, slow, losing information at every handoff. For small companies, a true repeatable system often never gets built at all.

And the human glue fails in ways everyone has learned to accept:

Not a discipline problem — a structural one. A human cannot answer every lead in under a minute, follow up six times on every thread, and run a live demo at 3am for the prospect in Lisbon.

Here's the scenario.

It's Tuesday, 6:40 in the morning. A founder with an exceptional product opens her laptop — not to build. To sell.

• Forty minutes building a list.
• An hour writing cold emails.
• The ad account — spend up, replies flat, no idea why.
• Two intro calls that go nowhere, retelling the pitch she's told two hundred times.
• The follow-ups she owes from last week.
• The CRM, updated at 9pm, from memory.

To grow past this, she's told to hire a marketer, an SDR, a salesperson, an agency, an ops person, and contractors. Not because the work is impossible — because no single system has been able to do it.

One intelligent system that replaces the fragmented machinery of customer acquisition.

She researches the market. She identifies likely buyers. She creates and launches campaigns. She writes outreach. She monitors target accounts. She responds to prospects. She runs live, Zoom-like sales calls. She answers questions. She follows up. She qualifies opportunities. She learns which messages are working, sees which prospects advance — and, most importantly, learns which prospects ultimately become customers.

And she works like a hire, not a tool: you manage her entirely by email. No dashboard to maintain, no CRM workflow to keep current. You email her instructions; she does the work; she emails you back with what happened.

There is deliberately no CRM integration, because there is no CRM — Sally is an all-in-one system and her pipeline is the system of record. The CRM is legacy software, built on the premise that humans do the selling and someone has to monitor and manage their activity. Take the humans out of the loop and what you actually want is answers: email her "give me a Q4 forecast based on historicals" or "who's likely to close in the next three weeks" and the answer comes back the way it would from a great head of sales.

You also decide exactly where her job ends. Want her to run one qualifying call and hand every deal straight to you — no follow-up, no back-and-forth after? She'll do that. But the most powerful version is the full cycle, start to finish: every call, every follow-up, every objection, all the way through the close.

Most outreach is timed for the sender. Sally's is timed for the buyer.

She doesn't work a list once and move on. She watches the accounts you care about — and the people on them — month after month, and moves when something changes.

1. A new hire lands in a role you sell to — she sees the title, and the introduction goes out that week, not at the next campaign.
2. Jim is promoted from manager to regional manager — she congratulates him, and the relationship is warmer the next time there's something to sell.
3. Jim leaves for a VP role at another company in the same market — she congratulates him there too. One relationship just became a second account, with someone who already knows you.

You teach her the triggers that matter in your market — reach out whenever an account hires a new VP — and she watches for them across every account, acts on them without being asked, and tells you what she saw and what she did in the same evening email.

Your week has changed shape. Every evening, one email.

Straightforward deals she closes herself, with your own payment link. Deals that need you arrive as a structured Deal Brief. On the last day of the month she asks where the handoffs landed; you answer in a sentence, she books the outcomes. You spend your time on the conversations only you can have.

Traditional software is built around isolated activities. Sally is built around revenue.

She doesn't merely know an ad received a click. She knows whether the person who clicked later took a meeting, what they cared about, which objections appeared, whether the opportunity progressed — and whether it became revenue. That creates a continuous learning loop across the entire customer journey.

1. She discovers that a particular pain point produces more qualified conversations.
2. She learns that one kind of company converts more often than another.
3. She recognizes that a message generating fewer clicks may still produce better customers.
4. She hears the same objection on sales calls and adjusts future advertising and outreach to address it earlier.

The campaign improves because the sales conversations improve it. The conversations improve because the campaign attracts better prospects. The entire system improves because it can see what actually turns into revenue. Today, humans perform this coordination. Sally turns that coordination into software.

Underneath sits something no ad platform or email tool has: with Stripe and QuickBooks connected — Chargebee, Recurly and others to follow — she sees the full lifecycle, from "never heard of you" through closed-won, contract signed, invoice paid. A verifiable outcome, and to our knowledge something no AI sales or marketing tool has built. So she can tell you that dentists who see one particular image and call to action click through and ultimately buy, while physical-therapy practice administrators click and reply — but stall, and exactly what they say in the sales conversations when they do. And when the pattern points at the product itself — a gap she keeps hearing about on calls — she bubbles it up to you intelligently, so you can weave it back into what you build.

Every night she runs an internal review — which lines got real traction, which fell flat, what worked and what didn't — and builds the winners into the next day's calls, ads, and outreach. Like a great salesperson replaying the day on the drive home, except she never skips a night and never forgets the lesson.

One wall never moves: outcomes teach her how to sell, never what's true. Product facts and pricing only ever change through you — and every change she makes to her own memory is cited, versioned, and reversible with one reply. When a prospect asks something she doesn't know, she has a hard rule: never invent an answer. She tells them she'll find out, asks you, follows up — and files your answer into her corpus of knowledge, so it's asked once and answered forever.

That is the loop inside one customer's account. But the loop that matters most to an investor is the one that runs across all of them.

Every call makes every Sally better.

This is the part I'd underline if you only read one section. It's why this is a company and not a good tool.

A sales conversation is the richest training signal in commerce, and it has always been thrown away. It happens once, in one room, in front of one person, and whatever was learned in it stays inside that one person's head until they leave for another job and take it with them.

Play that forward. Hundreds of sales conversations a day, then thousands, all feeding one system that learns from all of them. There is a point on that curve where Sally is running more sales conversations than any sales organization on earth has ever run, and learning from every single one — more deeply, more consistently, and more fluidly than any of them ever could.

Past that critical mass, using anything else to sell for you becomes the irrational choice. Not because of a feature. Because the thing on the other side of the table has had more reps than your entire industry combined.

And this is the kind of advantage that cannot be bought late. An incumbent with a hundred times our capital can ship an agent tomorrow; they cannot ship the conversations. The gap is cumulative, and it only opens in one direction. That is the road from a hundred customers to thousands — and the reason the thousandth customer is easier to win than the tenth.

Ads Sally generated.

Creative Sally produced for OpenVia, the founder's own access-control company — the copy, the design, and the per-platform formatting, shown as they run in-feed.

Not another AI SDR. Not a spam cannon.

An AI SDR books meetings a human still has to take, from sequences a human still has to write, into a funnel a human still has to run. It's one more tool on the pile. And the blast tools? Ten thousand name-swapped emails a day is exactly the machinery buyers have learned to delete.

Sally is not a writing tool, not a chatbot on a website, not another dashboard. She is the operating system for acquiring customers — the world's first comprehensive system for go-to-market: campaigns, outreach, conversations, live sales calls, qualification, follow-up, and the close, coordinated by one intelligence that learns from revenue.

Even her cold outreach proves the point: before any cold email goes out, she researches that prospect's company on the web and writes the whole email individually around what the company actually does — then an honesty check rejects name-swap templates, invented details, and manufactured urgency.

Who buys this? It's sales-motion dependent, not industry dependent.

Our target market is exclusively B2B — a uniquely tangible sale, and that isn't changing in the near term — and non-regulated industries, so not finance, banking, or medical, where licensure requirements apply. Beyond that, the screen is simple: do you sell B2B, mostly over email and Zoom? Then Sally can likely run your motion. She's not for the enterprise, where deals are done over steak dinners and golf — she's for the transactional end of the spectrum.

Our beachhead: founder-led B2B companies with contract values of $25K and under, where the founder is still doing the selling and trying to stop being the only salesperson. It's a genuinely painful point in time — and 95% of founders aren't great at sales, our own included. They're evaluating Sally against hiring their first or second rep; she's the alternative that costs a fraction and starts immediately.

The second core audience: companies selling into the US from abroad. For them the problem is structural — time zones that put every demo at 8pm local, an availability gap, and often a language barrier. Sally erases all three: she works every timezone and speaks any language, tailoring her outreach to prospects who speak Portuguese, Spanish, French — whatever it may be — in their native tongue.

That's already many hundreds of thousands of businesses — software companies, agencies, service providers — a huge market in itself. And the Stripe playbook applies from there: start with startups and small businesses, make money as some of them become big businesses, and work steadily up-market into larger accounts you never could have started with.

Nobody believes this from a deck. So we don't sell it from one.

The obstacle in front of us right now is not price, and it is not features. It's belief. We are at a specific and temporary moment: CEOs and founders want this to work. They can feel what it would mean for their business. They are simply not convinced it actually will — not for them, not with their product, not against their buyers.

That is not an objection you can argue someone out of. A better pitch doesn't move it. A case study about somebody else's company doesn't move it. The only thing that moves it is watching it work on their own market, with their own prospects, in front of them.

So we buy the proof. We take the loss at the front of the funnel and put real deals in a stranger's pipeline before they've paid us anything.

Step one: she pitches them their own company.

When a prospect shows real interest, Sally goes to their website, ingests everything publicly available about their company, and generates a complete sample deck and live sales presentation — for their product, with their positioning and their branding. Then she offers to present it.

"Hey Jim, one of the best proof points is exactly what I can do. Here is me selling your product."

So the buying moment isn't "watch Sally sell Sally." It's watching Sally sell their own product back to them, before they've signed anything. It lands 85–90% right — close enough that a founder polishes rather than rebuilds. We are not aware of a single competitor selling itself this way.

It costs us about $15 in inference, deliberately, because our first foot out the door has to be exceptional and that still takes the more expensive models. At a skeptic's 1% hit rate that's a $1,500 cost of acquisition — but she only builds them for prospects already showing interest, with budget and authority, so we expect closer to 10%, and we'll learn the true number at scale.

That buys a real conversation. It doesn't yet buy conviction, because a great pitch about your own company is still a pitch. So there's a second step.

The pilot has one job, and it isn't to impress. It's to produce deals — genuine prospects in genuine conversations, meetings on their calendar, opportunities they can see and touch and follow up on themselves. Theory converts nobody. A pipeline with their competitors' customers in it converts almost everybody.

And it inverts the close entirely. We never have to talk a skeptic into a $3,499 monthly commitment on the strength of a promise. We ask a founder who is currently receiving qualified meetings whether they'd like that to stop. It's the cheapest possible way to buy the only thing that's actually scarce right now, which is conviction.

The arithmetic holds. A pilot capped at a thousand dollars of real spend, converting into a Growth plan at $3,499 a month plus commission on everything she closes, pays itself back inside the first month and every month after that is margin. Even at a punishing conversion rate this is the cheapest customer acquisition in the category — and the customers it produces have already seen revenue from the product before their first invoice, which is the best churn insurance there is.

It's disciplined, not indiscriminate: pilots go to qualified prospects only — right motion, real budget, actual authority to buy — and every one is metered and capped. We are deliberately buying belief while belief is still the expensive thing. It won't be for long. In two years this market will take autonomous selling as read, and the companies with the conversations, the case studies and the compounding will already be chosen.

Sally is paid like a salesperson. Her incentives are her customer's.

She is compensated the way a salesperson is: a base that covers the acquisition motion itself, plus a cut of what she actually brings in. A customer who hires her and closes nothing pays us very little. A customer she makes rich pays us handsomely, and is delighted to.

The base scales with pipeline, not seats. Commission and ad management ride on top of every tier, so what we earn moves with what the customer earns.

Every tier gets every capability. The commission is the part that matters: most of what we make only exists once the customer has made money first.

Commission attribution is deliberately fail-closed: with the customer's permission the platform reads their own Stripe or QuickBooks — read-only — and only an airtight match between a closed deal and Sally's pipeline bills automatically. Uncertain matches ask first; no match is never billed; refunds claw back pro-rata. Trust in the billing is part of the product.

The margins underneath are software margins — 80 to 90 percent gross. We built and tested on the most capable frontier models from Anthropic and OpenAI, then progressively migrated the intelligence layer to far more economical open-source models — GLM 5.2 carries most of it today — with no quality loss we can measure. Ad-creative generation still runs on leading-edge models while the quality bar is the product; as each piece bakes, it moves down the same cost curve, and fine-tuning models of our own, custom-tailored to these use cases, is on the near-term list.

The comparison a buyer actually makes: an SDR runs ~$85K fully loaded and a demand-gen marketer ~$115K — and both draw that base regardless of results. Even hired outside the US, a human rep's monthly salary meaningfully exceeds the Growth plan. Sally's base is a fraction of one hire, works every timezone, never ramps, never churns, and most of her upside is earned only when revenue lands — and the customer also sheds thousands to tens of thousands a year in tooling and infrastructure, the CRM seats and point solutions she makes unnecessary. Software margins on a line item companies already accept as a percent of revenue.

Side by side: the way it's always been done, or Sally.

Price the year you hire your first sales rep and point a 5,000-person funnel at the market — the tools, the people, and the glue in between.

We didn't test Sally on a toy. We pointed her at ourselves.

The system is live in production today. Sally's first customer is SallySells — she runs her own go-to-market, works her own pipeline, and the sales call on our website is her, selling herself.

Her second is a beta customer in proptech, a business nothing like ours — exactly the point of a platform built multi-tenant from day one — actively using the product with great early results: a $5K deal already attributed to Sally's first month of effort.

The product works. What remains isn't intelligence. It's the harness.

The intelligence is rented. The harness is ours.

Everyone gets the same frontier models. They are a commodity we buy by the token, and they get cheaper and better every quarter whether we do anything or not. Being smart is not the moat, and it isn't the hard part any more.

The hard part is everything between a smart model and a salesperson you can actually hire. We call that the harness, and it is the difference between a demo and a business.

A founder will hand her a box of coat hangers.

Nobody's data is clean. Nobody exports neatly. What actually arrives is a PDF of conference attendees, a calendar export, a CRM dump with three tabs and a do-not-contact list buried in the third, a spreadsheet somebody merged cells in back in 2021, a screenshot of a LinkedIn search, and a sentence of context in the email body.

A good human assistant handles that without being told how. They open it, work out what it is, notice the tab that matters, ask one clarifying question if they need to, and get on with it. That is the bar, and it is a much higher bar than answering questions well. Sally has to meet it — she has to sort through the pile the way a person does, and then act on it.

The second one looks like one instruction and is roughly nine. Parse the file. Work out which of four hundred events were sales calls and which were standups, school pickups and dentist appointments. Identify the counterparty on each. Cross-reference against what actually closed. Find the ones that went quiet. Then write to each of them in a way that acknowledges a year has passed without being strange about it — and pace the whole thing so it doesn't land as a blast.

That is the work. Handling the mess is not a feature around the edges of the product; for the founder we're selling to, it is the product, because the mess is the honest state of their business.

And she needs more ways to reach people.

The rest of the harness is surface area: more lead sources feeding the top, more sending capacity underneath, and more channels than email at the point where a deal actually needs saving.

These are the classic high-touch human moves, and the reason they work is that they're rare and they cost something. A human rep can run them for three accounts. Sally can run them for every account that genuinely warrants one, fired by the actual state of the pipeline rather than by memory and good intentions — and she'll know afterwards whether the doughnuts closed the deal, because she can see all the way to the invoice.

That's the whole ambition of the harness: everything a human salesperson can do, at a scale no human salesperson can reach. It's unglamorous engineering, it isn't waiting on a model that doesn't exist yet, and it's exactly the kind of work that gets finished by a founder who isn't splitting his week.

This transition has already happened once. It was worth $3.6 billion.

Intercom sold software that helped human support teams answer customers — inbox, ticketing, knowledge base. Humans still read the questions, found the answers, resolved the tickets. Then it built Fin — and Fin crossed the boundary: it didn't help reps resolve tickets faster. It resolved them itself. Fin became the company's central product, gave the company its name, and led to an announced Salesforce acquisition valued at roughly $3.6B.

Almost everything sold as AI for revenue teams today sits in the middle band — copilots, assistants, AI SDRs. They make the human faster at operating the pile. That's a real improvement and a transitional one, and it is still priced against software budgets, because a human is still doing the job.

The lesson isn't the exit — it's the budget. Fin stopped competing for customer-service software spend and started competing for the vastly larger pool of money companies spend employing people to answer questions. Sally applies the same transition to revenue: a company stops buying software to assemble and manage a revenue operation — it buys the output of the revenue operation.

And the incumbents are structurally conflicted. Salesforce and HubSpot will ship agents — but agents designed to strengthen the CRM, because their economics depend on humans doing revenue work: more employees, more seats, more workflows. Kodak had the technical capability to build digital cameras. Sally is free to build the agent that makes much of the traditional CRM unnecessary. That freedom matters more than their resources.

$50K to finish the harness and take the market. Today isn't soon enough.

The instrument  Post-money SAFE · $7.5M cap · 20% discount

The money does two jobs: finish the production harness, and fund the pilots that buy our first customers their conviction. Five customers at our blended ~$2,500 average is $12.5K of monthly recurring revenue — default alive, operations and growth funded from revenue, with commission riding on top; ten takes it to ~$25K. At up to a thousand dollars of real spend per pilot, this round buys a lot of proof. The window is now, and it doesn't take much capital to seize it.

This is not a raise to find the product. The product exists, in production, selling. This is a raise to stop it waiting.

And what's waiting is specific. There are companies I could put on a pilot tomorrow — the right motion, real budget, people who have already told me they'd take it. I can't make them the offer today, because every pilot spends real inference and real sending infrastructure before it earns a dollar back. Funded, we go from one external company running Sally to five or more within days.

And each one teaches her something none of the others can — a different market, a different buyer, a different pile of mess handed over in a spreadsheet. That learning doesn't stay in one account. It goes straight back into the harness, where it compounds for everyone who comes after.

The fastest diligence takes eight minutes and no scheduling: take Sally's sales call — she'll pitch you the company herself, handle your objections, and follow up in the morning. Then reply to the email that brought you here.

Will we raise again? Maybe — and that's the point.

This is a market with real competition, and what we're building will inspire some who aren't competing with us today to start. A reasonable war chest might be the smart play. But there's a decent chance we never need one: a business growing 10–20% month over month at 80–90% gross margins can fund its own growth.

Here's the thing to understand about future fundraising: hit this round's stated goal — five to ten customers by end of year, each one a case study and a proof point — and we have optionality. Raise at a premium valuation, or just let the machine work and keep the company lean. We won't raise for the sake of raising money. We'll raise only when we know it buys acceleration.

And make no mistake — we're not raising to fund small ambitions. I'm expecting this to be a generational company. That said, dilution is a fair question for any early check — so the simulator below includes the follow-on round. Set its size and valuation yourself and watch what it does to your stake.

Here's what your investment could be worth. Break the assumptions yourself.

AI-native agent companies — software that does the work instead of assisting it — are commanding premium revenue multiples right now, and we expect that to hold for at least the next 12–24 months; Fin's ~$3.6B is the loudest example. Pick the multiple you believe, set the operating assumptions, and watch what a check becomes. Every number below is yours to change.