How to sell Jira and Confluence data to AI labs.
Who buys Jira tickets and Confluence pages, what buyers have paid, whether the data is yours to license, and how your admin exports and cleans a copy. Checked 19 November 2026. If your company has 30 or more office employees, Allied arranges a direct license with an AI lab and gets you a written offer.
The short answer
Yes. A company can license a copy of its Jira and Confluence history to an AI lab, or to a firm that supplies labs with training data. It grants a written license to a cleaned copy. It keeps ownership, and the originals stay in Atlassian.
Atlassian’s contract does not stand in the way. Its Customer Agreement says: “Customer owns all intellectual property and other rights in Customer Data.” Customer Data covers what your users put into the cloud products, and the agreement points to Atlassian’s documentation for how to retrieve it.1
Buyers name these records outright. Turing’s Project Lazarus asked companies for private codebases and “the operational artifacts around them (Jira tickets, PRDs, architecture docs, support threads).” It priced those artifacts at “$50 per artifact when bundled with a codebase.”2
Three things decide whether a sale can happen. The records have to be the company’s to license, not a client’s. Names, customer details and secrets have to come out of the copy. And someone with authority has to sign. The rest of this page takes them in order, then covers price, timing and what to do if the company is closing.
Every system Allied covers is on the sell page. If your company has 30 or more office employees, you can request an offer now.
| What sits in Atlassian | Where it is | Why a lab wants it | What your team removes first |
|---|---|---|---|
| Issues, comments and change history | Jira projects; the cloud backup carries work items, fields and comments6 | A problem, the attempts to fix it and the outcome, in order | Names, emails, customer names pasted into tickets |
| Sprints, boards and roadmaps | Jira board, sprint and Advanced Roadmaps data6 | How a team planned work and what slipped | Personal notes in sprint goals |
| Workflows and field schemes | Jira space configuration6 | The rules a team actually ran on | Little; check custom field names |
| Specs, runbooks and decision pages | Confluence pages and whiteboards7 | Why a system or process works the way it does | Hostnames, IP addresses, credentials pasted into pages |
| Service desk requests | Jira Service Management | A customer’s question and how it was resolved | Customer names, emails, phone numbers, account IDs |
| Attachments | Jira and Confluence file storage, if selected in the backup67 | Screenshots, logs and drafts behind a ticket or page | Most of it; files are the hardest part to clean |
Who buys Jira and Confluence data
The buyers are AI labs and the firms that build training data for them. They want the working record of a company: how a bug was filed, argued over and closed, and how a spec changed before launch. Public web pages rarely show that.
micro1 lists software and engineering among the functions it licenses data from: “Source code documentation, engineering workflows, bug tracking, product planning, and technical knowledge bases.” It looks for established companies with 30+ full-time employees and says payments typically range from $100K to over $3M.3
Mercor’s company data page lists Confluence among the 100+ supported tools a company can connect, next to Slack, Gmail and Salesforce. It says payment is issued by wire once the anonymized dataset reaches the buyer, which typically occurs within 2-4 weeks of the signed agreement.4
Shutdown firms sell this kind of record too. MIT Technology Review reported that 47% of SimpleClosure deals now include code or data sales. Buyers it names include foundation AI labs and vendors that build simulated workspaces for AI training.5
For the full list of buyers, see who buys AI training data. For every record type labs want, see what data AI companies buy.
What makes a Jira or Confluence archive worth more
A lab training an agent to do project work needs the steps, not just the result. A ticket that was opened, reassigned, argued over in comments, linked to a spec page and finally closed with a fix is one complete example. A ticket closed with no comment is close to worthless.
Buyers price that context. Turing’s launch post paid “$1,000 per PR, up to $10K per codebase” for modern code, and “$50 per artifact” for tickets, specs and docs bundled with a codebase. So Jira and Confluence history is worth most next to the code and chat it refers to.2
Sunset, a shutdown firm, told MIT Technology Review it values a client’s assets by head count, years in operation, volume of data and how the team used its tools. A team that ran its work through Jira and wrote its decisions in Confluence scores well on the last one.5
What lowers value: projects that were bulk-closed during a cleanup, spaces that were mostly templates, and wikis that stopped being updated years ago. What raises it: long history, linked issues and pages, real disagreement in comments, and outcomes you can check, such as a release that shipped or a bug that stayed fixed.
Allied’s own figure: old operational business data could be worth $100K–$2M. Jira and Confluence on their own sit at the low end of that. Licensed with Slack, email, CRM or code, they add the context a buyer pays for. More on price is in how much company data is worth.
Is the data the company’s to license
Usually yes, with exceptions your counsel checks before any offer is accepted.
Atlassian’s side is settled by its contract: as between the parties, the customer owns its Customer Data, and Atlassian keeps the rights in its products, templates, formats and dashboards. That means the content your team wrote is yours, while Atlassian’s own templates and software are not part of what you license.1
Client work is the common exception. An agency or consultancy that ran client projects in its Jira, or kept client deliverables in Confluence, usually did so under contracts that make that material confidential or the client’s. Those projects and spaces stay out unless the client agrees in writing.
Service desk data needs its own look. Jira Service Management requests come from your customers. Read your customer contracts and privacy policy before including them. The rules for that are on the support ticket page.
Employee content is the third check. Comments and pages carry names, opinions and sometimes personal details. See employee consent for AI training data and whether it is legal to sell company data.
How your admin exports a copy
Your own Atlassian admin makes the export, inside your own account. Allied never sees, reads or receives it. Allied sees only a description of the projects and spaces that you approve.
Jira Cloud has a backup manager under Settings, System, Import and Export. Atlassian’s documentation says the backup carries work items and field content, board and sprint data, comments, user and group settings, space configuration and Advanced Roadmaps. It does not carry automation flows or third-party app data. Backups that include attachments need 48 hours between runs, and Atlassian recommends staying within a total data size of 25 GB.6
Confluence Cloud’s backup manager sits under Settings, Data management. The backup holds pages, whiteboards, team calendar data and, if selected, attachments, including those in the archive or trash. Spaces in the trash are not backed up. After the backup completes, the download link is available for 14 days, and only one backup file is stored at a time.7
For a buyer, a full site backup is often more than you want to send. Many companies export only the projects and spaces in scope, as CSV or XML from Jira and HTML or XML from Confluence, and clean that subset. Agree the format with the buyer before anyone exports anything.
Self-managed Jira or Confluence Data Center works the same way in principle. Your admin exports from your own servers, and nothing leaves your network until a cleaned sample is approved.
Removing personal data before anything is licensed
Jira and Confluence hold more personal data than most teams expect. Assignee and reporter fields, @mentions, watchers, comments, page authors and edit history all name people. Service desk tickets name customers. Pages and attachments often hold pasted logs, credentials, hostnames and screenshots.
Your team de-identifies the copy in your own systems. Replace names and emails with consistent placeholders, so a buyer can still see that the same person reported and later closed a ticket. Strip customer names, account numbers and contact details from ticket text. Rotate any password or token found in a page or comment before the copy leaves, because removing it from the copy does nothing for the live system.
Attachments are the hard part. A screenshot can show a customer’s screen, and a log file can hold IP addresses and tokens. Many sellers leave attachments out of the first license, or include only file types their team can scan.
Buyers say they scrub too. MIT Technology Review reported that suppliers employ data scientists to scrub personally identifiable information and exclude HR records, and that buyers say they scrub as well. Do your own first anyway: a buyer’s cleaning happens after the data has left you.5
A sample of the cleaned copy is approved before the full copy goes straight to the buyer. For the method in detail, see how to anonymize company data for AI training and how data handling works.
The license, the timeline and payment
What you sign is a data license: the records in scope, what the buyer may do with them, the de-identification standard, the price and when it is paid. The terms worth reading closely are covered in AI data licensing agreement terms.
Buyer timing varies. micro1 says it has 30 days to review data after transfer, then pays within 30 days of the invoice. Mercor says payment typically occurs within 2-4 weeks of the signed agreement.34
On a direct license arranged by Allied, a written offer naming the lab follows one to two weeks after you sign with Allied. Payment arrives by wire 30 to 60 days after the lab accepts delivery. The license is non-exclusive, so the same records can be licensed again.
Most of the elapsed time is your own team’s: confirming scope, exporting and cleaning a sample. A company that already has a clean export moves fastest.
If the company is closing: do not cancel Atlassian first
Closing companies are where most of these sales happen, and the Atlassian subscription is often one of the first bills cut. That order can destroy the asset.
Atlassian’s knowledge base warns: if you unsubscribe from all Jira products on your site, your Jira data will be deleted immediately.9
For a cancelled subscription, Atlassian says the site stays accessible for 15 more days, then is deactivated and enters a data retention period. On Free, Standard, Premium and Enterprise plans that period is 60 days. At the end, the data is permanently deleted and cannot be recovered. Its Customer Agreement also lets either party end the agreement if the other ceases operation without a successor.81
So keep the subscription and an admin account until the export is confirmed. Make the backup, download it before the link expires, and check that it opens. Then decide what to license.
The sale itself has to be signed by someone with authority while the company still exists, or by an assignee or trustee afterwards. See selling startup data when shutting down, the shutdown checklist, what happens to data in an assignment for the benefit of creditors and how long to keep business records after closing.
How Allied works on a Jira and Confluence license
Allied arranges the license. You license directly to the AI lab. Allied never sees, reads or receives your data. It sees only a description of the projects and spaces that you approve.
On a discovery call, you describe the systems you use, how many years of history they hold, and what stays out. If the records fit what a lab is buying, Allied takes the description to the labs and comes back with a written offer naming the lab, one to two weeks after you sign with Allied.
Your team then exports and cleans a sample, the lab checks it, and the full copy goes straight from your systems to the lab. The license is non-exclusive. Allied works with companies that have 30 or more office employees. To start, request an offer.
Questions about selling Jira and Confluence data
Can I sell my company’s Jira and Confluence data to AI companies?
Yes, as a license to a cleaned copy. Atlassian’s Customer Agreement says the customer owns its Customer Data. AI lab suppliers such as Turing, micro1 and Mercor ask for tickets, specs, bug tracking and knowledge bases. The records must be the company’s, with personal data removed, and someone with authority signs.
Who buys company data for AI training?
AI labs and the firms that build training data for them, such as micro1, Mercor and Turing. Shutdown firms such as SimpleClosure and Sunset also sell closing companies’ records to labs. Allied arranges direct licenses between a company and an AI lab.
How much can a company get for its Jira and Confluence history?
There is no public price for Jira or Confluence alone. Buyers pay most when tickets and pages come with the code and chat they refer to; the published buyer prices are in the section above. Allied’s figure for old operational business data is $100K–$2M.
How do AI labs pay companies for training data?
By wire, after the data is delivered and the buyer has reviewed it. On a direct license arranged by Allied, payment arrives 30 to 60 days after the lab accepts delivery.
How long does it take to sell company data to an AI lab?
On a license arranged by Allied, a written offer follows one to two weeks after you sign with Allied. Most of the rest is your own team exporting and cleaning a sample, so a company with a clean export moves fastest.
How is Jira and Confluence data anonymized before it is licensed?
Your own team does it on a copy in your own systems. Names and emails become consistent placeholders, customer details come out of ticket text, credentials are rotated and removed, and attachments are scanned or left out. A sample is approved before the full copy goes to the buyer.
What are the risks of selling company data to AI companies?
Personal data left in comments or attachments, client material the company does not own, and customer contracts that forbid sharing. Read the license terms on use, resale and deletion closely. Cleaning the copy yourself, before it leaves, is the main safeguard.
Does Allied see our Jira or Confluence content?
No. Allied never sees, reads or receives your data. Your admin exports and cleans the copy in your own systems, and it goes straight to the lab. Allied sees only a description of the projects and spaces that you approve.
What happens to our Jira data if we cancel Atlassian while shutting down?
Atlassian says unsubscribing from all Jira products on a site deletes the Jira data immediately. A cancelled paid site enters a retention period, then the data is permanently deleted. Export and check the backup before cancelling anything.
Can a company that is going out of business still sell its data?
Yes, while someone has authority to sign: the company’s officers before dissolution, or an assignee or trustee afterwards. The records must survive until then, so keep the subscriptions running until the export is confirmed.
I’m retiring and closing my business. Is our Jira and Confluence data worth anything?
Possibly, if the company has years of real project history and 30 or more office employees. Value rises when Jira and Confluence are licensed with chat, email, CRM or code. A discovery call with Allied answers it for your records.
Do we lose our data if we license it?
No. A license grants the buyer the right to use a copy. You keep ownership and the originals. Allied’s licenses are non-exclusive, so the same records can be licensed again.
Are there alternatives to selling data directly to an AI lab?
Yes. Data suppliers such as micro1 and Mercor buy records and build training sets, and shutdown firms sell records as part of a wind-down. A direct license with the lab is the other route, which is the one Allied arranges.
Sources
- Atlassian Customer Agreement, effective 1 October 2026, sections 4.4, 12.4 and 13
- Turing, “AGI Advance” newsletter 57, Project Lazarus launch, 25 March 2026
- micro1, “Monetize your company’s operational data” (data partnerships page), read 19 November 2026
- Mercor, “Monetize your data” (company data page), read 19 November 2026
- Matthew King, “AI labs are buying data from dead startups to train their models,” MIT Technology Review, 13 October 2026 (reporting)
- Atlassian Support, “Export data from Jira Cloud”
- Atlassian Support, “Export a data backup file of your instance” (Confluence Cloud site backup)
- Atlassian Support, “Track storage and move data across apps” (subscription end and data retention)
- Atlassian Support, “What happens to Jira Cloud projects when you unsubscribe from a product”
This page is general information, not legal advice. Whether your company can license its Jira and Confluence records depends on your client and customer contracts, your privacy policy and the law where your employees and customers are. Atlassian’s terms and buyer programs change; read the current versions. Ask your own counsel.
Find out what your project history is worth.
A few questions. About two minutes. How Allied works.