Who takes chat transcripts for AI training
5 of 12 organisations we read pay the owning business for a licence to its data, read 28 October 2026.
The way it buys
Does it ask for chat logs
What it wants, in a sentence of its own
A seller's first step
Who keeps the rights, quoted
Each organisation, answer by answer
Each row is one organisation. Click an answer to see the sentence behind it, quoted from the page with the date we read it. Where it says "not stated", the page we read is silent on that question.
| Organisation | The way it buys | Does it ask for chat logs | What it wants, in a sentence of its own | A seller's first step | Who keeps the rights, quoted |
|---|---|---|---|---|---|
| Appen | builds and sells a conversation corpus of its own | asks for text of any kind | not statedNot addressed in Off-the-shelf AI training datasets | not statedNot addressed in Off-the-shelf AI training datasets | All are collected under documented consent and licensing terms, with provenance records… |
| AWS Data Exchange | lists owners' data for buyers to purchase | asks for company records, chat logs not named | your data files, tables, and APIs | Become a data provider Make it easy to reach AWS customers | not statedNot addressed in Data providers |
| Bitext | builds and sells a conversation corpus of its own | asks for chats, dialogues or support conversations | not statedNot addressed in Hybrid datasets to fine-tune LLMs | not statedNot addressed in Hybrid datasets to fine-tune LLMs | not statedNot addressed in Hybrid datasets to fine-tune LLMs |
| Defined.ai | pays the owning business for a licence | asks for company records, chat logs not named | We’re actively looking for partners with high-demand data types. | Our in-house experts will guide you along the way in four easy steps. | Assessment of your licensing rights as well as legal, compliance and cybersecurity checks |
| Getly | lists owners' data for buyers to purchase | asks for chats, dialogues or support conversations | not statedNot addressed in Marketplace home page | Explore products arrow_forward Start selling | not statedNot addressed in Marketplace home page |
| Grepped | lists owners' data for buyers to purchase | asks for chats, dialogues or support conversations | Docs, contracts, reports, slides, and PDFs: the written record of what an organization… | Answer a few questions about your company and see an estimated payout range in 2 minutes. | Your data stays under your control until a buyer purchases. |
| micro1 | pays the owning business for a licence | asks for chats, dialogues or support conversations | We are looking for data that reflects how work is actually performed within an… | We identify licensable datasets across your business and tailor an offer based on the… | Companies retain ownership of their underlying data. |
| Nexdata | builds and sells a conversation corpus of its own | asks for chats, dialogues or support conversations | not statedNot addressed in LLM training datasets | not statedNot addressed in LLM training datasets | not statedNot addressed in LLM training datasets |
| Nyne | pays the owning business for a licence | asks for chats, dialogues or support conversations | 3+ years of real operational workflows that your teams already run on. | Build a detailed inventory of your tools, data sizes and history. | Every package stays within the scope you approve. |
| Protege | pays the owning business for a licence | asks for company records, chat logs not named | The Series A extension follows rapid adoption across healthcare, media, audio, motion… | Contact contact@withprotege.ai | while maintaining clear rights protections and provenance |
| Shaip | builds and sells a conversation corpus of its own | asks for chats, dialogues or support conversations | not statedNot addressed in Licensed text datasets | not statedNot addressed in Licensed text datasets | not statedNot addressed in Licensed text datasets |
| Troveo | pays the owning business for a licence | asks for chats, dialogues or support conversations | Tickets, sprints, wikis, and complete project histories | Get an estimate in minutes, with a standard agreement ready when you are. | Thousands of companies trust Troveo to license their data, protect their rights, and pay… |
Where each answer was read (60)
- Appen, Off-the-shelf AI training datasets, read 2026-10-08, The way it buys:
Off-the-shelf (OTS) AI training datasets are pre-collected, pre-licensed datasets that teams can license and use immediately
- Appen, Off-the-shelf AI training datasets, read 2026-10-08, Does it ask for chat logs:
from reinforcement learning (RL) tasks and verifiers to speech, code, and enterprise data
- Appen, Off-the-shelf AI training datasets, read 2026-10-08, What it wants, in a sentence of its own: Nothing on this page answers it (26,258 characters of page text, read 2026-10-08).
- Appen, Off-the-shelf AI training datasets, read 2026-10-08, A seller's first step: Nothing on this page answers it (26,258 characters of page text, read 2026-10-08).
- Appen, Off-the-shelf AI training datasets, read 2026-10-08, Who keeps the rights, quoted:
All are collected under documented consent and licensing terms, with provenance records your legal and compliance teams can review.
- AWS Data Exchange, Data providers, read 2026-10-27, The way it buys:
Make it easy to reach AWS customers with your data files, tables, and APIs on the AWS Data Exchange catalog
- AWS Data Exchange, Data providers, read 2026-10-27, Does it ask for chat logs:
make the world's third-party data easy to find in one data catalog
- AWS Data Exchange, Data providers, read 2026-10-27, What it wants, in a sentence of its own:
your data files, tables, and APIs
- AWS Data Exchange, Data providers, read 2026-10-27, A seller's first step:
Become a data provider Make it easy to reach AWS customers
- AWS Data Exchange, Data providers, read 2026-10-27, Who keeps the rights, quoted: Nothing on this page answers it (6,647 characters of page text, read 2026-10-27).
- Bitext, Hybrid datasets to fine-tune LLMs, read 2026-10-28, The way it buys:
Bitext has unveiled a dataset that’s not just vast, with nearly 27,000 rows, but also meticulously curated for customer support applications.
- Bitext, Hybrid datasets to fine-tune LLMs, read 2026-10-28, Does it ask for chat logs:
meticulously curated for customer support applications
- Bitext, Hybrid datasets to fine-tune LLMs, read 2026-10-28, What it wants, in a sentence of its own: Nothing on this page answers it (5,354 characters of page text, read 2026-10-28).
- Bitext, Hybrid datasets to fine-tune LLMs, read 2026-10-28, A seller's first step: Nothing on this page answers it (5,354 characters of page text, read 2026-10-28).
- Bitext, Hybrid datasets to fine-tune LLMs, read 2026-10-28, Who keeps the rights, quoted: Nothing on this page answers it (5,354 characters of page text, read 2026-10-28).
- Defined.ai, partner programme page, read 2026-10-27, The way it buys:
Defined.ai helps you turn that data into meaningful new sources of income
- Defined.ai, partner programme page, read 2026-10-27, Does it ask for chat logs:
Through your business operations (customer interactions, stock information, shipping routes etc.) you generate data
- Defined.ai, partner programme page, read 2026-10-27, What it wants, in a sentence of its own:
We’re actively looking for partners with high-demand data types.
- Defined.ai, partner programme page, read 2026-10-27, A seller's first step:
Our in-house experts will guide you along the way in four easy steps.
- Defined.ai, partner programme page, read 2026-10-27, Who keeps the rights, quoted:
Assessment of your licensing rights as well as legal, compliance and cybersecurity checks
- Getly, Marketplace home page, read 2026-10-28, The way it buys:
Templates, design assets, music, courses, AI prompts - yours to download the moment you pay. Creators keep 80–90%
- Getly, Marketplace home page, read 2026-10-28, Does it ask for chat logs:
Synthetic B2B SaaS Support Dialogues — 484 Conversations (JSONL)
- Getly, Marketplace home page, read 2026-10-28, What it wants, in a sentence of its own: Nothing on this page answers it (10,575 characters of page text, read 2026-10-28).
- Getly, Marketplace home page, read 2026-10-28, A seller's first step:
Explore products arrow_forward Start selling
- Getly, Marketplace home page, read 2026-10-28, Who keeps the rights, quoted: Nothing on this page answers it (10,575 characters of page text, read 2026-10-28).
- Grepped, business seller page, read 2026-10-27, The way it buys:
Approved listings reach AI labs. When they purchase, the money comes to you.
- Grepped, business seller page, read 2026-10-27, Does it ask for chat logs:
Chats, threads, and channels where people coordinate work, including the decisions they reach and the reasoning behind them.
- Grepped, business seller page, read 2026-10-27, What it wants, in a sentence of its own:
Docs, contracts, reports, slides, and PDFs: the written record of what an organization knows and how it works.
- Grepped, business seller page, read 2026-10-27, A seller's first step:
Answer a few questions about your company and see an estimated payout range in 2 minutes.
- Grepped, business seller page, read 2026-10-27, Who keeps the rights, quoted:
Your data stays under your control until a buyer purchases.
- micro1, operational data programme, read 2026-10-27, The way it buys:
We license data generated through your company’s everyday operations.
- micro1, operational data programme, read 2026-10-27, Does it ask for chat logs:
Support documentation, ticket workflows, QA processes, customer operations, and knowledge management
- micro1, operational data programme, read 2026-10-27, What it wants, in a sentence of its own:
We are looking for data that reflects how work is actually performed within an organization.
- micro1, operational data programme, read 2026-10-27, A seller's first step:
We identify licensable datasets across your business and tailor an offer based on the nature, scale, and value of your data.
- micro1, operational data programme, read 2026-10-27, Who keeps the rights, quoted:
Companies retain ownership of their underlying data.
- Nexdata, LLM training datasets, read 2026-10-28, The way it buys:
Nexdata provides high-quality LLM datasets for dialogue systems, instruction tuning, and multilingual AI training
- Nexdata, LLM training datasets, read 2026-10-28, Does it ask for chat logs:
Nexdata provides high-quality LLM datasets for dialogue systems
- Nexdata, LLM training datasets, read 2026-10-28, What it wants, in a sentence of its own: Nothing on this page answers it (10,290 characters of page text, read 2026-10-28).
- Nexdata, LLM training datasets, read 2026-10-28, A seller's first step: Nothing on this page answers it (10,290 characters of page text, read 2026-10-28).
- Nexdata, LLM training datasets, read 2026-10-28, Who keeps the rights, quoted: Nothing on this page answers it (10,290 characters of page text, read 2026-10-28).
- Nyne, partnership programme page, read 2026-10-27, The way it buys:
Nyne turns your de-identified operational workflows into frontier-grade training data, under your governance and on your terms.
- Nyne, partnership programme page, read 2026-10-27, Does it ask for chat logs:
Slack · Microsoft Teams · Google Chat
- Nyne, partnership programme page, read 2026-10-27, What it wants, in a sentence of its own:
3+ years of real operational workflows that your teams already run on.
- Nyne, partnership programme page, read 2026-10-27, A seller's first step:
Build a detailed inventory of your tools, data sizes and history.
- Nyne, partnership programme page, read 2026-10-27, Who keeps the rights, quoted:
Every package stays within the scope you approve.
- Protege, Home page, read 2026-10-08, The way it buys:
Monetize your existing data assets or content with Protege
- Protege, Home page, read 2026-10-08, Does it ask for chat logs:
Protege streamlines access to private and proprietary real-world data across domains.
- Protege, Home page, read 2026-10-08, What it wants, in a sentence of its own:
The Series A extension follows rapid adoption across healthcare, media, audio, motion capture, and more
- Protege, Home page, read 2026-10-08, A seller's first step:
Contact contact@withprotege.ai
- Protege, Home page, read 2026-10-08, Who keeps the rights, quoted:
while maintaining clear rights protections and provenance
- Shaip, Licensed text datasets, read 2026-10-28, The way it buys:
Shaip’s text data catalog gives AI teams ready-to-license text datasets for training, fine-tuning and evaluating large language models
- Shaip, Licensed text datasets, read 2026-10-28, Does it ask for chat logs:
conversational Q&A, code, news, clinical, academic and multilingual data
- Shaip, Licensed text datasets, read 2026-10-28, What it wants, in a sentence of its own: Nothing on this page answers it (6,160 characters of page text, read 2026-10-28).
- Shaip, Licensed text datasets, read 2026-10-28, A seller's first step: Nothing on this page answers it (6,160 characters of page text, read 2026-10-28).
- Shaip, Licensed text datasets, read 2026-10-28, Who keeps the rights, quoted: Nothing on this page answers it (6,160 characters of page text, read 2026-10-28).
- Troveo, company data revenue page, read 2026-10-27, The way it buys:
AI labs and startups are actively paying six figures for company data exports.
- Troveo, company data revenue page, read 2026-10-27, Does it ask for chat logs:
Email threads with senders, recipients, and linked responses
- Troveo, company data revenue page, read 2026-10-27, What it wants, in a sentence of its own:
Tickets, sprints, wikis, and complete project histories
- Troveo, company data revenue page, read 2026-10-27, A seller's first step:
Get an estimate in minutes, with a standard agreement ready when you are.
- Troveo, company data revenue page, read 2026-10-27, Who keeps the rights, quoted:
Thousands of companies trust Troveo to license their data, protect their rights, and pay them on time.
Read and excluded
4 named documents fell outside what this board reads once we had read them; their reasons follow. They are tallied, never scored as answers.
- Defined.ai, open-source datasets for conversational AI: a buyer-facing article on licensed versus open-source conversational datasets; Defined.ai's page for owners is recorded as source 1
- Human Native, home: the read holds Human Native's home page shell only, with no sentence on what it takes or how
- Unidata, datasets: Unidata's dataset index, a navigation list and contact form with no sentence on conversation text
- Snowflake Marketplace: the Snowflake Marketplace page for consumers of data; it names no door for a business offering its own data
How we read and count
Each organisation's own data partnership, text dataset and data marketplace pages were read through estate-fetch and stored whole (sha256 on every row).
For every page, five questions were answered from its own words: the way it buys, does it ask for chat logs, what it wants, in a sentence of its own, a seller's first step, who keeps the rights, quoted.
Every stated answer carries the sentence it rests on, checked against the stored text; 'not stated' means the page read does not address the question.
A organisation with several pages reads 'varies' where they disagree.
12 organisations, 12 pages.
16 sources, with links
- Defined.ai, Partnership Programs (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Defined.ai, open-source datasets for conversational AI (read, set aside)
- micro1, Monetize your company's operational data (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Grepped for Business (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Troveo, Turn your company data into revenue (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Nyne, Data partnerships (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Protege, home (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Human Native, home (read, set aside)
- Appen, off-the-shelf datasets (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- AWS Data Exchange, data providers (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Shaip, licensed text datasets (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Unidata, datasets (read, set aside)
- Nexdata, LLM training datasets (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Snowflake Marketplace (read, set aside)
- Getly, marketplace for digital products (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
- Bitext, hybrid datasets to fine-tune LLMs (quoted for reporting from the publisher's data partnership, text dataset and data marketplace pages)
Cite it or embed it
5 of 12 organisations we read pay the owning business for a licence to its data, read 28 October 2026.
The way it buys, by organisation: pays the owning business for a licence 5 of 12, builds and sells a conversation corpus of its own 4 of 12, lists owners' data for buyers to purchase 3 of 12 (12 organisations, read 28 October 2026).
Does it ask for chat logs, by organisation: asks for chats, dialogues or support conversations 8 of 12, asks for company records, chat logs not named 3 of 12, asks for text of any kind 1 of 12 (12 organisations, read 28 October 2026).
What it wants, in a sentence of its own, by organisation: stated 7 of 12, not stated 5 of 12 (12 organisations, read 28 October 2026).
A seller's first step, by organisation: stated 8 of 12, not stated 4 of 12 (12 organisations, read 28 October 2026).
Who keeps the rights, quoted, by organisation: stated 7 of 12, not stated 5 of 12 (12 organisations, read 28 October 2026).
Suggested citation: "Who takes chat transcripts for AI training", updated 2026-10-28, https://sellchattranscriptstoai.com/datasets/chat-transcript-buyers/.
Get the CSV (60 answers, worked out 2026-10-28)