Tools for AI Data Sellers: Buyer Comparison, Program Terms and Estimates

DataPacts seller tools are dated reference tables that show what AI data buyers publish about their programs: who they accept, what they claim to pay, which license terms they use and what their own estimators return. Four tools cover data acquired for AI training, and every value is the stating party's claim, checked on November 9, 2026.

The tools sit within the reference on AI training data licensing. None of them promises a price, ranks buyers by payout or referral fee, or asks a seller to upload data.

DataPacts Editorial Team checked every value on this page against the publisher's own page on November 9, 2026; the page on how DataPacts sources, dates and checks every figure sets out the rules.

What the seller tools do, and what none of them promises

The tools answer one question each from public sources, and none of them quotes, predicts or promises a price for a seller's data. The four kinds of output are listed below.

• Eligibility rules: the headcount, history, country and decision-maker conditions buyers state.
• Payout claims: the ranges, tiers and paid-out totals buyers state about their own programs.
• License terms: the exclusivity, ownership, deletion, acceptance and payment clauses buyers publish.
• Estimator outputs: the indicative ranges that calculators of buyers and intermediaries return.

Buyers' figures are marketing, and they do not always agree on the same page. Handshake AI's data partnerships page carries the hero line "Earn $100K to $4M+", while its estimator is headed "Your data could be worth up to $100,000" (both checked 2026-11-09). Troveo states "$50M+" paid to creators on its content-owners page and "$20M+" paid to licensors on its business-data page, both checked 2026-11-09. DataPacts shows both sides.

One editorial rule governs every estimate: a range appears only when a dated public source carries it, and always as the publisher's own claim, never as a valuation.

The four seller tools at a glance

DataPacts runs four seller tools, and only the program terms table is published as of November 9, 2026. The table compares them along the seller's path by input, output and source basis, as of 2026-11-09.

How each tool works: inputs, outputs and sources

Each tool below is described by the same three attributes: what the seller enters, what the tool returns and where the values come from.

Payouts and terms recorded as each buyer's dated claim

The terms table records each program's payout claim, payment timing, exclusivity and referral fee exactly as the buyer's page states them, with the date the page was checked. Sellers filter by buyer type and data type. micro1 states three payout tiers, checked 2026-11-09: USD 100K to 500K for qualified enterprise data, USD 500K to 3M for large-scale or ongoing participation, and USD 3M+ for highly unique proprietary data. Mode states that a company's data "could be worth" USD 100K to 5M, with its calculator labeled "Indicative only" (checked 2026-11-09).

Mercor states that it pays by wire on delivery of the anonymized dataset, typically 2 to 4 weeks after signing, and covers extraction, anonymization and transfer costs (checked 2026-11-09). Handshake AI claims "$250M paid out to partners" and "1K+ partner companies" without defining a partner, both checked 2026-11-09, and the row keeps that scope note. Troveo's two totals sit in one row as two claims. Each company data partnership program in the table also has its own entry among the company data partnership programs.

Matching a company profile to published eligibility rules

The buyer match compares a company profile with the eligibility rules buyers publish and shows which rules the profile meets, misses or that a buyer does not state. The seller enters the four inputs listed below.

• Headcount band: full-time employees, now or at peak.
• Country: where the company and its records are based.
• Company status: operating, defunct or insolvent.
• Data type: the records on offer, such as email, tickets, CRM entries or code.

Published rules differ, among them the following, all checked 2026-11-09. micro1 names "established companies with 30+ full-time employees" and data primarily in English. Nyne states 20+ full-time U.S. W-2 employees, 3+ years of operational workflows and a clear decision-maker. Handshake AI's data partnerships page states no minimum, while its license and referral pages state 20+ full-time employees and 3+ years in operation. Aba Compute, a broker, states 20+ full-time employees, 3+ years of operating history and one decision-maker who can sign.

A rule absent from a buyer's page reads "not stated", and no output says a buyer will take the data. Company status decides the most for a shut-down or defunct company: Turing's Project Lazarus intake names active, winding-down and wound-down companies, and Labelbox states that wound-down companies are eligible for Prism (both checked 2026-11-09). Selling data from a shut-down or defunct company has its own page.

Comparing contract clauses across buyers' published terms

The clause comparison lines up what each buyer's published terms say about exclusivity, ownership, deletion, acceptance and payment, and marks terms a buyer has not published. Its five clauses are listed below.

• Exclusivity: exclusive or nonexclusive grant, and for how long.
• Ownership of derivatives: who owns models and outputs built on the data.
• Deletion and retraining: when raw data is deleted and whether trained models change.
• Acceptance: how long the buyer has to reject a delivery.
• Payment trigger: signing, delivery, acceptance or each later sale.

Published terms differ sharply, as of 2026-11-09. micro1's Data License Master Terms, last updated 2026-11-08, take a nonexclusive, irrevocable and perpetual license (clause 1.2), give the licensee ownership of derivatives and models (2.2) and deem a delivery accepted unless rejected within 30 business days (1.3). Labelbox's Prism FAQ makes licenses exclusive by default, provides a kill clause forcing deletion within 30 days and states that trained models cannot be unlearned. Turing's Project Lazarus licenses non-exclusively, with Turing the only seller for the first 90 days, and pays a revenue share with nothing upfront.

Handshake AI publishes no exclusivity, deletion, retraining or indemnity terms, so its row reads "not published". What each clause in AI data licensing agreements means is set out on the page on AI data licensing agreements, clause by clause.

Estimates shown only as the publisher's own claim

A buyer's estimator is marketing that returns an indicative range for that buyer's own program, so DataPacts shows estimator outputs side by side only as each publisher's dated claim and never as a valuation. The seller enters one profile made of the inputs each estimator asks for; Handshake AI's asks for employees, tools, years and country (USA, Canada, Europe or Other), checked 2026-11-09.

Estimator outputs contradict their own publishers, all checked 2026-11-09. micro1's estimator shows an example output of USD 50,284 to 92,776 without stating its inputs, below the floor of micro1's own lowest tier. Handshake AI's page source holds, besides the heading quoted above, a second state of the same widget reading "$400,000-$800,000+" (checked 2026-11-09). Nyne's estimator returned four ranges for the same default inputs (100 employees, 10 years, USA) on four page loads on 2026-11-09: USD 51,996 to 121,745, 56,037 to 131,207, 41,420 to 96,982 and 38,632 to 90,454.

Turing's Project Lazarus page shows four unnamed calculators applied to one made-up company, with ranges of USD 197K to 394K, 322K to 689K, 269K to 537K and 600K to 800K (checked 2026-11-09), and no range can be attributed to a buyer. Intermediaries publish estimators too, among them Aba Compute, a broker (USD 300K to 800K), Avelence, a seller-side adviser (USD 160K to 500K for 20 employees, 10 years, US), and Pewter, a broker (USD 385,000 to 895,000 for 100 staff and 10 years), all checked 2026-11-09; each range is that intermediary's claim. An estimate does not say what company data is worth to AI companies, and the valuation methods are set out under what company data is worth to AI companies.

How publishers label their own estimators

Each of the eight estimators in the table carries a label on its own page saying that the output is indicative, illustrative, a scenario or not an offer, and the table quotes each label. The eight, alphabetical as of 2026-11-09, are among the published estimators; Turing's comparison of unnamed calculators is not an estimator of its own.

How the tools are built: sources, check dates and no fees

Every value in a DataPacts tool comes from a primary source, carries the date it was checked and is attributed to the party that states it. The rules are set out in how we source, date and check every figure.

Where a buyer's own pages disagree, both claims appear side by side with their source pages. Rows follow alphabetical order or another stated neutral criterion, such as data type, and never payout or referral fee. micro1 states a company referral fee of USD 50,000 for deals up to USD 700,000 and 7.5% above, and USD 75,000 up to USD 1M for defunct companies with 7.5% above (checked 2026-11-09); its rows still sit where the alphabet puts them.

DataPacts charges the company nothing for an enquiry or introduction, takes no cut of any sale and has joined no referral program, so the tools carry no referral links. Any program it joins later appears on the page on how DataPacts is funded and as a one-line note on the affected page.

Benefits, risks, downsides and alternatives of using seller tools, including not selling

The tools shorten the reading of buyers' pages, but they reflect what buyers say rather than what sellers were paid. The three benefits are listed below.

• One dated table: the claims of every listed program in one place.
• Visible contradictions: conflicting figures on one buyer's pages appear side by side.
• Early eligibility gaps: missed published rules show before any contact.

The three risks are listed below.

• Changing claims: micro1 updated its master terms on 2026-11-08, one day before the check.
• Misread matches: meeting published rules can read as an acceptance, which it is not.
• Anchoring estimates: a range can anchor a negotiation, as micro1's and Nyne's outputs show.

The two downsides are listed below.

• Claims only: the tables record what buyers state, not amounts sellers confirm they received.
• Blank cells: unpublished terms, such as Handshake AI's license terms, stay empty.

The four alternatives are listed below.

• Reading the buyer's own program page and terms directly.
• Engaging an independent adviser or counsel.
• Requesting an independent introduction through the company data enquiry.
• Not selling or licensing at all.

Whether to sell or license company data to AI companies is the holder's decision, and keeping the archive is a valid result. The steps from readiness to payment are set out in the guide on how to sell or license company data to AI companies.

Where the tools lead next: buyers, prices, law and the seller path

The tools answer what buyers publish; the pages below answer who the buyers are, what data has sold for, which law applies and in which order a sale runs. Six related pages are listed below.

• Who buys AI training data: every buyer and program, compared without fees.
• AI training data prices: what buyers pay per work, hour, image, token and record.
• AI training data law by jurisdiction: copyright, privacy, employees and contracts.
• Buyer program changelog: every dated change to payouts and terms.
• Evaluating an offer for your data: price, payment trigger, acceptance and license terms.
• Alternatives to selling data, including not selling at all.

Frequently asked questions

Is there a calculator that tells me what my company data is worth?

No public calculator values a specific company's data; buyers' estimators show indicative ranges for their own programs. micro1's example output falls below its own lowest tier, and Nyne's changed between four page loads on 2026-11-09.

Do the tools rank buyers by payout or referral fee?

No; rows are ordered alphabetically and never by payout or referral fee. micro1's referral fee, quoted above, does not move its rows.

Buyers' referral terms are compared under referral programs of AI data buyers.

Which AI data buyers accept small companies?

Published minimums differ: micro1 states 30 or more full-time employees, while Handshake AI's license and referral pages and Nyne state 20 or more and three years. These are among the rules checked on 2026-11-09.

Are the payout figures confirmed payments?

No; the figures are buyers' claims with check dates, not payments that a seller has confirmed.

Advertised and confirmed amounts are compared under reported payouts to data sellers.

Where is the data sale readiness checklist?

The readiness checklist is a seller-process guide, not a tool.

The checklist sits on the page on data sale readiness.

Do I have to upload data to use the tools?

No; the tools read only buyers' public pages and hold no seller data.

Preparing an archive before any transfer is covered under PII removal and de-identification.