Handshake AI Data Partnerships: Payouts, Rules and Terms

Handshake AI's data partnerships license a company's internal business data, the records in the tools it already runs on, and its own page says it helps companies do that with $100K to $4M+ payouts. It looks for companies with 20+ full-time employees and 3+ years in operation, prices a company's data within 12 to 24 hours of the details it needs, and pays on delivery milestones written into each agreement, with the whole process typically taking 7 to 14 business days. Handshake is contractually restricted to using the data for AI model training and evaluation, the company keeps ownership, and Handshake removes the personal and business identifiers itself before it uses anything. What its pages do not publish is the licence itself: how long it runs, whether it is exclusive and whether Handshake may pass the data on.

This guide is for the founder, CEO, CFO or counsel of a company that has heard Handshake AI pays for company data and wants to know what it actually buys, what it pays, whether the company qualifies and what it would be signing. It reads Handshake's own pages and the press about it, section by section, in the order a company meets them: what Handshake is and why it buys, what it takes, its published payouts and its calculator, who it takes and from where, its timeline and how it pays, the licence, what happens to personal data, and how to apply or be introduced. It ends with a short comparison against Turing's Project Lazarus, micro1 and Scale AI, the programmes a company usually weighs beside it, and a worked example.

AIData.Broker is in Handshake AI's referral programme, as it is in Mercor's, Turing's, micro1's and Scale AI's. We introduce a holder to every buyer whose published rules take its data, free for the holder, because the buyer pays us its published referral payout out of its own money: for Handshake, a flat amount for each company that completes its data transfer, shown on Handshake's page here, and the holder sees the most any buyer may pay us before we introduce it. Where a sale run by us earns more, and only for holders based in the United States, the United Kingdom and Canada, we act as the holder's agent under a mandate and the holder pays our fee only if the sale closes; every fitting partner is introduced then as well. Either way the holder itself licenses its data to the buyer, and we do not buy it, take ownership of it, or license or sell it in our own name. Every term below is Handshake's own published wording or named press, cited to its record; where a line is our reading, it says so.

This guide is research from public sources, not legal, tax or investment advice.

Contents

1. What Handshake AI is, and why it buys company data
2. What Handshake AI buys
3. What it pays: the published payouts and the calculator
4. Who it takes
5. The process and its timeline
6. How and when it pays
7. The licence: what is published and what is not
8. What happens to personal data
9. Applying directly or being introduced
10. Handshake beside Turing, micro1 and Scale AI
11. Worked example: a field-service contractor
12. Questions people ask
13. What this means for you

What Handshake AI is, and why it buys company data

Handshake AI began as a careers network that connects students and graduates with employers. Its AI division, Handshake AI, sells AI labs expert human model validation, RL environments and evaluations, drawing on a network of graduate-level experts who write tasks, answers and judgments for models. That business is large and moving fast: its founder told TIME it grew to more than $1 billion in gross annualized revenue in 16 months, working with eight frontier AI labs, and its own job posts describe paying ~$60M a month to over 30,000 experts and managing lab accounts at $100M+ run rate. It bought Cleanlab, a start-up that audits data labels, in January 2026. All of these figures are the company's own or reported from it, not audited accounts.

The data partnerships are a newer line beside that. Labs are now training models to work inside companies, and a model that has to handle an invoice dispute, a hiring decision or a project handover learns that from records of real work, which no public website holds. Handshake's FAQ says licensed data is used "to train and evaluate private AI models. This typically includes creating tasks that test and improve how models perform real-world work." It also says which records it values most: datasets that reflect "real organizational decision-making", naming hiring logic, business process flows and document-level communications. Other buyers say the same in nearly the same words: their stated reason for paying a company is that public and synthetic data cannot replicate how an organisation actually works, and that labs pay for what is not on the open web.

Handshake is also going out to find that data rather than waiting for applications. A job post live on 15 October 2026, since closed, described an outbound team going directly to companies to license their proprietary data, naming workflow data, code and support transcripts, which the team then cleans, rights-clears and hands to labs. So a company may hear from Handshake before it has applied, and the questions in this guide apply either way.

Two facts set the frame. First, Handshake is a principal: it licenses your data for its own account and turns it into training material and evaluations it sells to labs. Its published figure is the opening of its own negotiation, not a reading of what labs would pay for your records. Second, the programme is young: no holder has yet confirmed in public a sale of company records to it, so its scale rests on its own statements, which the next sections read closely.

What Handshake AI buys

Handshake buys operational data: the email, chat, documents, accounts and project records a business produces by running, as they sit in the software it already uses. It does not ask a company to create anything. Its licence page lists AutoCAD, Dropbox, Figma, Gmail, Google Drive, Outlook, Monday.com, NetSuite, Notion, Paychex, QuickBooks, ServiceTitan, Slack and 1K+ more as examples of the sources it buys. That list covers almost every function of a small or mid-sized company: correspondence and chat, shared files, accounting and payroll, design, project management, and field service.

The list matters because each system on it brings its own rights questions. A Slack export holds employees' messages; QuickBooks and NetSuite hold customers' and suppliers' accounts; Paychex holds payroll, which is employees' pay, tax and bank details; ServiceTitan holds customers' names, addresses and job notes. The software vendor's own terms also decide what you may export and pass on. Before you share a sample from any of them, read our pages for the systems you use, such as Slack, Google Workspace, Microsoft 365, QuickBooks and NetSuite, and the guides to selling Slack data, email archives and financial records.

What raises the value, in Handshake's words

Handshake's FAQ names four key factors behind the value of a company's data: "the complexity of your business operations and how well your data captures them, the years of operational history available, the relevance of your industry and data to AI training, and the size of your business, including your number of full-time employees." Two of those, history and size, are what its calculator asks. The other two, complexity and relevance, are what its team judges once it sees the records, and they are where one company of a given size differs most from another.

In practice that means three things raise a company's figure. Linked records are worth more than isolated ones: a support ticket tied to the email thread, the invoice and the fix shows work moving start to finish, which is what buyers say they want; five buyers of company records, Handshake among them, name scheduling, dispatch and field-service systems for exactly that reason. More systems are worth more than one, which Handshake's own calculator now counts. And decisions in writing (why a candidate was hired, why a bid was priced as it was) are worth more than records of routine activity, because that is the reasoning a model cannot learn elsewhere.

It also buys from people, not only companies

Handshake reaches a company's knowledge through individuals too. Business Insider reported that it offered people $6 per accepted page, up to $30,000, for professional documents they own, and Columbia Journalism Review published an offer it made a freelance journalist of a thousand dollars per case of reporting notes. Neither is a company partnership. Both are a reason for an employer to have a written rule on whether staff may license work files, which the guide to employee emails and chats in AI training covers from the employer's side.

What it pays: the published payouts and the calculator

Handshake publishes more figures than most buyers, in three places: a range on its pages, a total it says it has paid, and a calculator. Each answers a different question, and none of them is an offer.

The published range

The licence page says Handshake helps companies license their data with $100K to $4M+ payouts, and its main data-partnerships page headlines the same $100K to $4M+. Its FAQ is more careful: payments to date have ranged from $100K to $4M, "depending on the scope of the partnership". Read together, the range describes deals Handshake says it has done, from the smallest to the largest, not a band your company will land in. The same FAQ calls the calculator a preliminary estimate and says the full value "will be more accurately assessed during scoping", with the price agreed in writing before any data moves.

The licence page also shows $250M paid out to partners and 1K+ partner companies. That figure gives no period and does not say whether "partners" means companies that licensed data or includes the experts Handshake pays for its other work, so it is not evidence of how much it has paid for company records. Its own disclaimer adds that any payout "will vary based on several factors", among them "the outcome of our due diligence process and your acceptance of our data licensing terms and conditions" ( licence page). We could not find those terms and conditions published.

The calculator

The calculator on the data-partnerships page asks four things: the number of employees, the number of tools the company uses, years in business and its country. Between 17 and 25 October 2026 Handshake replaced its old engine with one of its own, and the outputs we read on 25 October, for a US company using five tools, are these:

Three things stand out. First, the calculator's figure for a small company sits at or below the floor of the published range: a 20-person company sees a ceiling of $100,000, where the range starts. So the range's bottom is not a promise of a minimum, and a small company should expect its figure to come from the calculator's end of the scale. Second, the calculator stops growing at its sliders' top, so it cannot say anything about a large company. Third, it is lower than it was. Until mid-October Handshake's estimator ran the same formula as micro1's and Nyne's: a fixed $140,000, plus $2,333 a person and $11,667 a year of records. The new engine prints less for the same company, and Handshake has not said why. A buyer can rewrite its estimator overnight, which is the plainest sign that an estimator measures the buyer's opening position, not your data.

Below three years in business the calculator gives no figure and answers that the team should be contacted. For how all the buyers' estimators are built, and why the same company sees figures about twenty times apart from different buyers, read how AI data buyers calculate offers. For a reading that does not come from a buyer, value your data: it prices your records against published prices and sales, and the guide to what company data is worth explains how.

Where the company is based changes the figure

The calculator's own settings, in the page's script, price each country choice with a factor: USA 1.00, Canada 0.75, Europe 0.75, Other 0.30. The outputs follow: the same 100-person, 10-year company is $193,000 to $385,000 in Europe and $77,000 to $154,000 in the rest of the world. micro1 and Nyne use the same location weights. Handshake does not say why. The likely reasons are language, the law that governs the records and how easily a buyer can check the rights, which the next section takes up.

Who it takes

Handshake's rules are short and, read closely, soft. They cover size and age, and say little about place or industry. The words matter, because a rule a buyer states as a condition rules a company out, while a rule it states as what "typically" fits leaves the company's case open.

Size and age

The licence page says: "We're looking for companies with 20+ full time employees & 3+ years in operation." That reads like a threshold. The FAQ, answering "What companies qualify?", softens it: companies with those numbers "are typically more likely to be a fit", Handshake looks for "internal business data that captures how real work gets done, such as documents, communications, and records of business processes", and "Eligibility depends on the type, volume, and usability of the data."

So, by its own words, the size and age rule is a screen rather than a gate, and Handshake's team decides each case at review. What it takes can also change faster than its page, as every buyer's rules have this year. The screen does decide the first impression: the calculator starts at 20 people and three years, and a company below either is not ruled out by the words but should expect to make its case on what its records contain, and should ask before it prepares anything. Full-time employees is the count Handshake names; a company that relies on contractors should say so at the first conversation rather than counting them in.

Where the company is run from

No Handshake page rules out a country. The data-partnerships page asks which country the company is in, offering USA, Canada, Europe and Other, its sign-up form asks for the company's country with no stated restriction, and the FAQ's answer on who qualifies names no country. But the calculator's factors above show where its demand sits: the US first, Canada and Europe next, everywhere else far behind. Its payment terms state no currency, method or payout country.

Our reading: a company headquartered and run from the US is the case Handshake is built for. A company run from elsewhere is not ruled out by any published word, but it should ask at the first contact whether Handshake is taking companies from its country now, before it prepares or shares anything, and it should plan for the cross-border questions: a US buyer may withhold tax on a licence fee paid abroad (see US withholding tax on data licence payments), and the records may be governed by privacy law that a US buyer must also meet. The guides to selling data from outside the US and to the buyers that take international holders set out the alternatives.

Industry, and companies that have closed

Handshake's pages name no industry they exclude, and the sources they list run from design studios (Figma, AutoCAD) to accounting (QuickBooks, NetSuite) and field service (ServiceTitan). Relevance to AI training is one of its four value factors, so industry moves the price rather than the answer. No Handshake page says whether it takes a company that has stopped operating. A closing company should ask directly, and read the guide to who owns a dissolved company's data first, because the person entitled to sign changes when a company winds down.

The process and its timeline

Handshake's FAQ describes the steps in one paragraph: you share a few details about your business; its team reviews them for fit and schedules a conversation about the company and the data it may license; then it assesses the data's value, agrees scope and terms with you, and coordinates delivery and preparation. "Your agreement outlines the approved uses, de-identification requirements, and payment milestones." Its page breaks the middle into two stages: exploring the opportunity (reviewing the records, the history they cover and the work they capture, and agreeing what sensitive information must be excluded) and your team's approval of the proposed scope, protections and commercial terms before you commit.

The clock is one of the fastest any buyer publishes. Handshake says it can give an initial review and pricing within 12 to 24 hours of the details it needs, and that the full process typically takes 7 to 14 business days to finalise the transaction. It handles the transfer itself: its team works with you one to one, at no cost to you, with remote, in-person and physical transfer options, chosen by where the data is stored and what you agreed to share.

That clock measures Handshake's side. Yours is usually longer, because the questions a buyer's counsel asks take time to answer properly: our rights check lists 27 of them, 16 of which block a sale until they are answered, from customer contracts and software terms to employee messages and old privacy promises. A company that starts those before its first call can move at Handshake's speed; one that starts them after the price arrives will find the price waiting on its own lawyers. The readiness check works through them in order, and how to prepare data for sale covers the rest.

How and when it pays

Handshake publishes how it pays but not when, in days. Its FAQ says: "Payment terms are negotiated as part of your agreement and tied to defined delivery milestones. You'll receive written details on the payment amounts, milestones, and schedule before any data is transferred." The price itself is agreed in writing before any data is transferred.

Milestone payment is common for data deals that arrive in stages: a first package, then the rest, or a sample, then the full set. It protects the buyer from paying for data it has not received, and it protects the holder from delivering everything before any money moves. What it leaves open is what each milestone is. "Delivery" can mean the day you hand data over, the day Handshake finishes de-identifying it, or the day it accepts it as matching the agreed scope, and those can be weeks apart. Ask for each milestone to name the event that triggers it, the amount, the number of days to payment after it, and what happens if Handshake finds part of a package unusable.

The other large programmes pay on different events. Mercor pays by wire on delivery of the anonymised dataset to the buyer, typically 2 to 4 weeks after signing; micro1 lets you invoice only after it has accepted the data in writing; Turing pays from each sale of your record to a lab. Handshake's milestones sit between delivery and acceptance, and the agreement decides which. For every buyer's payment trigger and day count side by side, read when AI data buyers actually pay.

The licence: what is published and what is not

Handshake publishes the use, not the grant. Its FAQ says: "Handshake is contractually restricted to using the data for AI model training and evaluation only", and "Your company retains ownership of its data, and the scope, approved uses, protections, and compensation are agreed upon before any data is transferred." Before you commit it asks you to review who may receive the data, how it may be used, and the safeguards required, and says it partners with all of the leading frontier AI labs.

That is a narrower use than some buyers take, and it is worth having in writing. But the four terms that decide what a licence is worth are not on any Handshake page: how long it runs, whether it is exclusive, whether Handshake may sublicense (pass the data or the rights on to labs, and they to others), and what happens to models already trained if it ends. Its page also does not say who carries the legal risk of the transfer if a customer, an employee or a software vendor later objects. A missing term is not a generous one. It is a term the buyer will write in its own agreement, after you have shared a sample.

The nearest public evidence of how Handshake drafts a grant is a contract it sent a freelance journalist for her reporting notes, published by Columbia Journalism Review on 15 October 2026: a non-exclusive, royalty-free, fully paid up, perpetual, irrevocable, transferable and sublicensable licence. That was an individual's contract, not a company agreement, so it does not say what Handshake offers a company. It does show that a narrow use clause on a page and a broad, permanent, transferable grant can sit in the same company's paperwork, which is why you ask for the full agreement before you compare Handshake's price with anyone else's.

What each term is worth

The terms are not legal fine print; they are most of the price. On the market's evidence an exclusive licence fetches 2.0 to 5.8 times a non-exclusive one, so a holder asked for exclusivity without a separate price for it is giving away its largest lever. Disclosed AI licences with large platforms run two to three years or at least three years, and licensing counsel recommends pricing a perpetual right to models already trained separately from a finite licence to the data. A training-and-evaluation use is broad in practice (it covers the main reasons a lab buys data), while an evaluation-only licence, where data only tests models, fetches 0.06 to 0.39 times a full commercial one.

So the questions for Handshake are concrete. Is the licence perpetual, and if so what is the price of a fixed term with a paid renewal? Is it exclusive, for how long, and at what premium? May Handshake sublicense, and if so to whom, and may those recipients pass it on? Does "training and evaluation" cover building products sold to third parties? What happens to the raw copy and to trained models if the licence ends? The licence terms page sets out where a holder should stand on each, the guide to data licensing agreements walks a full agreement, and the offer check reads a written offer against the clauses buyers publish.

What happens to personal data

Handshake does the de-identification itself, after the data reaches it. Its FAQ says it "utilizes structured processes designed to remove direct and indirect business and personal identifiers from datasets prior to using them for training and evaluation of AI models", and adds, plainly, that those processes are designed to minimise re-identification, "however, no de-identification process eliminates that risk entirely." Before the transfer, it asks you to define which records are included and what sensitive information needs to be excluded, and to understand who will access and prepare the data. The licence page adds that "All data is scoped, de-identified, and confidential" ( licence page).

Two consequences follow for a holder. The first is that identifiable data leaves your systems: Handshake's process removes identifiers "prior to using" the data, which means after receiving it, so for a period Handshake holds your customers' and employees' details as they are. That is a choice, not a flaw, but it is one of the points on which buyers differ most. Scale AI says it anonymises inside your own systems before anything is extracted, Turing works in isolated, air-gapped processing after you approve the scope, and Mercor runs its own automated pipeline. Who de-identifies, where and to what standard is priced: in the Spirit Airlines auction one bidder offered 35% more for the same records if it could do the scrub itself ( the bids), and in that sale the price is paid only on delivery of de-identified data certified by an independent agent.

The second is that the legal duty stays with you. Handshake's de-identification protects Handshake's use; it does not answer whether you may hand the records over in the first place. Employees' messages are the question that most often holds a sale up: Spirit's own records sale hearing has been adjourned again and again while the estate negotiates with unions over employee information, most recently to 17 November 2026. Customer records raise the old promises in your privacy policy (see is it legal to sell customer data), and health records a separate federal standard (see HIPAA de-identification for AI training). Settle what you will exclude before the first call, and ask Handshake in writing where the raw copy is held, for how long, who can see it, to what standard it is de-identified, and whether an independent party checks the result. Our de-identification standard and the de-identification tool show what a holder can do before any buyer sees a file.

Applying directly or being introduced

There are two ways to reach Handshake's data partnerships team. Both lead to the same team, the same review and the same agreement, which you sign with Handshake.

Applying directly

The licence page has a short form, "Find out what yours is worth", that asks about you and your company and its country ( licence page). Handshake's team then reviews it for fit and, by its FAQ, follows up to schedule a conversation. Before you send it, have four things ready: a list of the systems you use and how many years each holds, a count of full-time staff now and at your largest, a short note of what you will not share (named clients, health data, a director's mailbox), and who in the company can sign. A company that applies itself saves the buyer whatever it budgets for an introducer; buyers set aside a share of a deal, or a fixed amount, for whoever brings them a company, and a holder that comes direct can ask for that budget in its price, though nothing Handshake publishes says it agrees to that.

Being introduced

Handshake runs a referral programme open to anyone. Its FAQ says there is no limit on how many companies one introducer may refer, that "you must be the first to introduce the company to us", and that "Companies that have already been referred or are already in contact with our team aren't eligible." It pays the introducer only once the company "becomes a Handshake AI data partner and completes its data transfer". So an introduction costs the company nothing, the company's own deal with Handshake is untouched by it, and a company already talking to Handshake simply goes on talking to it.

This is how an introduction through us works. Value your data first: if the rules Handshake applies take your company, the valuation says so and offers one button, Introduce me. Pressing it is your consent to the introductions it names, and we file each one in the partner's own referral form within one business day. Handshake's data partnerships team then writes to you itself, within one to two days by its own account, and deals with you directly from there: the review, the price, the agreement and the payment are between you and Handshake. The same press introduces you to every other partner whose rules take your data, which is the practical difference from applying to one buyer: several answers to compare, not one. Whether you go direct or through us, read going direct or through us for what each route costs.

Handshake beside Turing, micro1 and Scale AI

A company that has looked at Handshake has usually looked at three other programmes run by large suppliers of human data to labs. They publish different things, and the differences are in the terms more than in the headline figures. Mercor, the fifth, publishes a price for each kind of company it wants and has its own guide, as do Turing's Project Lazarus and micro1.

Turing no longer buys a record for one payment. Since rebuilding its site by 2 November 2026 it prepares a company's record, licenses it to labs and pays a revenue share on every licence sold, with nothing upfront, and if nothing sells you owe nothing. You keep ownership and give Turing the right to prepare and license the record; preparation costs are covered from sales, and your share of each sale is set in the agreement and not published. Against Handshake it trades a certain, earlier payment for the chance of several later ones, and it gives no figure until it has reviewed a sample within 5 business days.

micro1 publishes the clearest tiers and the broadest licence. Its master terms take a licence that is for ever, worldwide, paid in full once and irrevocable, which it may grant on to others, who may grant it on again, and since 30 October 2026 it owns everything made from the data, models and datasets included. It ranks US companies first, then other Western markets, and licenses data only to labs and enterprises in the US and allied countries. It asks you to promise you have not already licensed the same data on similar terms, which matters if you license to Handshake first.

Scale AI is the most careful about where the data is handled: it anonymises inside your systems before extraction and releases data package by package, each with your sign-off. It publishes an illustrative $10K to $1M+ that grows with how often data is delivered, and its own job post says it is still defining how these deals are priced. It prefers 40+ employees, so a company of 20 to 40 people fits Handshake's screen and not Scale's preference, though both say they decide case by case.

Two conclusions follow. Handshake is the fastest of the four to a price and the narrowest on use, but it publishes the least about its grant, so its agreement has to be read against micro1's published terms before the two prices can be compared. And the four are not exclusive of each other unless a licence makes them so: a company that licenses non-exclusively can license again, which is why the order you sign in, and each buyer's warranty about earlier licences, matter. The guide comparing every company-data buyer programme covers these four and a dozen more, and the buyer board keeps each buyer's live terms with the date they were read.

Worked example: a field-service contractor

Example. A heating, plumbing and electrical contractor in Ohio has 60 full-time staff, twelve years of history, and runs on ServiceTitan for jobs and dispatch, QuickBooks for accounts, Gmail and Google Drive for correspondence and files, and Slack for its technicians. An email arrives from a buyer's outbound team. Its founder wants to know whether Handshake is worth answering and what else is open.

Does it fit? On Handshake's published words, yes: it is above 20 full-time staff and three years, it is run from the US, and all five of its systems are among the sources the licence page names. Field-service records are also what buyers say they want: Handshake and Mode list ServiceTitan, and Scale names Field Services (HVAC, Plumbing, Electrical) among its industries, though it prefers 40 or more staff, which this company has. micro1's rules ( 30+ staff, English, US first) fit too, and The Information reported start-ups paying HVAC companies $150,000 for data in July 2026, a headline figure rather than a price list. Mercor publishes no listing for a home-service business ( its listings by kind of company), so it is the least likely fit.

What do the calculators say? None takes these exact inputs, so the contractor reads the nearest. Handshake's calculator, with five tools, shows $93,000 to $185,000 for 50 people and ten years, and $257,000 to $513,000 for 100. micro1's estimator showed $285,896 to $604,463 for 50 people and ten years when we read it in October. The gap between the two is not a judgment of the contractor's records; neither tool has seen them. An independent valuation of what it actually holds is the better starting point.

What must it settle first? ServiceTitan holds homeowners' names, addresses and the notes technicians write in their homes; QuickBooks holds customers' accounts; Slack holds staff messages. Before any sample leaves, the contractor decides what it will exclude (payment details, any job notes about a customer's health or home security), checks what its own privacy notice told customers, checks the software terms for each system, and tells staff what it is doing with their messages. Handshake will de-identify after receipt; Scale offers to do it inside the contractor's systems. That difference is worth asking each to price.

What does it ask Handshake? For the full agreement before any sample; the term, exclusivity and sublicensing in it; what event triggers each payment milestone and how many days after it the money arrives; who carries the risk if a customer later objects; and where the raw copy sits until it is de-identified. If it is also talking to micro1, it asks micro1 to narrow its warranty about earlier licences, so a non-exclusive deal with Handshake does not breach it. It may still sign with Handshake. It will know what it signed.

Questions people ask

What is the Handshake AI data partnership?

It is Handshake AI's programme to license a company's internal business data, the records in tools such as Gmail, Slack, Google Drive, NetSuite, QuickBooks and Figma, for training and evaluating AI models, with payouts its page puts at $100K to $4M+. Handshake is contractually restricted to training and evaluation, the company keeps ownership, and Handshake removes identifiers itself before use. It is part of the same company that runs the Handshake careers network.

How much does Handshake AI pay for company data?

Its FAQ says payments to date have ranged from $100K to $4M, depending on scope. Its calculator is lower for small companies: up to $100,000 for 20 people and three years, and $257,000 to $513,000 for a US company of 100 people with ten years and five tools. Both are indications; the price is set after Handshake's team reviews the data.

Is Handshake AI's data partnership legit?

Handshake is a large, independently reported business: its AI division reported more than $1 billion in gross annualized revenue in 16 months, and it publishes its process, its use restriction and its de-identification method. What no holder has done yet is confirm in public a sale of company records to it, so treat the programme as real but young, read its full agreement for the terms its pages leave out, and check any offer that claims to be from it against its own pages. SimpleClosure has warned founders about buyers advertising $1 million to $5 million offers with no completed purchase behind them.

Who qualifies for a Handshake AI data partnership?

Its licence page names 20+ full-time employees and 3+ years in operation, and no industry it excludes. Its FAQ softens that to "typically more likely to be a fit" and says eligibility "depends on the type, volume, and usability of the data". Below three years its calculator asks you to contact the team rather than giving a figure.

Does Handshake AI take companies outside the US?

No page rules a country out: its form asks for the country, offering USA, Canada, Europe and Other. But its calculator prices Canada and Europe at 0.75 and other countries at 0.30 of the US figure, so the US is plainly its first market. A company run from elsewhere should ask at the first contact whether Handshake is taking companies from its country now.

How long does Handshake AI take to pay?

It says it prices within 12 to 24 hours of the details it needs and finalises the deal in 7 to 14 business days, with payment tied to delivery milestones set in the agreement. It publishes no number of days from a milestone to the money, so ask for that in writing. Your own rights work usually takes longer than Handshake's process.

Can I license the same data to Handshake and another buyer?

Only if every licence you sign is non-exclusive and none forbids it. Handshake does not publish whether its licence is exclusive, so ask. micro1's licence is non-exclusive but asks you to promise no earlier licence on similar terms, and Turing's is non-exclusive after its first 90 days. Plan the order before you sign the first.

How do Handshake AI referrals work?

Anyone can refer a company through its referral programme, with no limit on how many. Only the first to introduce a company counts, a company already referred or already in contact with Handshake's team is not eligible, and the introducer is paid only once the company completes its data transfer. The company pays nothing for being introduced and signs its own agreement with Handshake.

What this means for you

Handshake AI is one of the most transparent buyers about its process and its speed, and one of the least transparent about its licence. That combination rewards a holder who prepares: the price arrives within a day, so everything that decides whether the price is good has to be settled before the first call.

1. Check the fit honestly. Handshake's screen is 20 full-time staff and three years, softened by its own FAQ. Run your size, place and systems through the buyer match tool to see which other buyers' published rules you meet, and read each one's live terms on the buyer board.
2. Get your own reading of value. One buyer's calculator is one anchor. Value your data against published prices and sales before you hear Handshake's number.
3. Settle the rights file first. Customers, employees and software terms are what hold sales up. Use the readiness check and the pages for your systems under rights.
4. Decide what you will exclude. Handshake asks you to define it; do it before a sample leaves, with the de-identification standard as your checklist.
5. Ask for the agreement, then compare. Get Handshake's full licence and put any written offer through the offer check; the negotiation page explains where to stand on term, exclusivity, sublicensing and payment.
6. Hear from more than one buyer. Apply to Handshake directly, or be introduced to it and every other buyer whose rules take your data with one press after your valuation. Either way, compare price, licence, de-identification and payment together, as the guide to every company-data buyer programme does.