How to assess engineering issue histories and review decisions for data licensing

Short answer

To assess engineering issue histories for referral to SourceX, look for companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia (outside the U.S., Canada and the EU, at least 70% of the data must be in English) that consistently document their internal engineering problems, solutions, and review processes. Records like these can justify a conversation; whether a current SourceX program wants them is checked in review.

Visual briefFull-size diagram
A decision tree asks who wrote the issues and whose account holds them, then sorts three Illustrative companies: A, written by employees in the company's own account, is described as company original; C, held in one person's account, waits until control is settled; B, written by customers or outsiders, is listed apart for rights review.
Illustrative. Companies A, B and C are the invented cases on this page; a label records authorship and control, never whether SourceX would accept the records.

Identifying Value in Engineering Issue Histories for AI Data Licensing

For AI developers, well-documented engineering issue histories and review decisions represent a rich, proprietary dataset. This guide helps referral partners understand what to look for in potential companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia (outside the U.S., Canada and the EU, at least 70% of the data must be in English) that may have such valuable operational data.

What are Engineering Issue Histories and Review Decisions?

This category of data refers to a company's internal records concerning problems encountered in product development, software engineering, or operational technology. It includes:

  • Issue Tracking: Logs of bugs, feature requests, incidents, and tasks.
  • Troubleshooting & Resolution: Detailed descriptions of problems, diagnostic steps, analyses, and implemented solutions.
  • Design & Code Reviews: Records of discussions, decisions, and changes made during review processes.
  • Retrospectives & Post-Mortems: Summaries of projects or incidents, identifying what went well, what didn't, and lessons learned.
  • Decision Logs: Documentation of choices made, the rationale behind them, and their outcomes.

These records often contain metadata, code snippets, diagnostic outputs, team discussions, and performance metrics, painting a comprehensive picture of how a company builds, maintains, and improves its technical operations.

Why is this Data Valuable for AI Developers?

AI developers seek this type of data for several reasons:

  • Pattern Recognition: AI models can learn from past issues and resolutions to predict future problems or suggest optimal solutions.
  • Process Improvement: Understanding how different decisions impact outcomes can help AI optimize workflows and decision-making processes.
  • Training & Development: Real-world engineering challenges and their resolutions are excellent for training AI that assists developers or automates debugging.
  • Risk Mitigation: Analyzing historical incidents can help AI identify potential failure points and develop preventative strategies.
  • Proprietary Insights: This data comes from each company's own operations, which is why it is not found in public datasets.

How to Identify Companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with Valuable Engineering Data

As a referral partner, you don't need to analyze the data itself. Instead, focus on indicators that a company in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE or Saudi Arabia systematically generates and stores this type of information. Look for companies that:

  • Utilize Robust Internal Tools: Companies using comprehensive issue tracking systems (e.g., Jira, Asana, GitHub Issues, internal wikis, or custom platforms) are likely to have structured records.
  • Practice Structured Development Methodologies: Agile, Scrum, DevOps, or similar methodologies often require detailed documentation of tasks, issues, and reviews.
  • Have a Culture of Documentation: Do they emphasize documenting decisions, post-mortems, and lessons learned? This indicates a higher likelihood of rich historical data.
  • Operate Complex Technical Products or Services: companies managing intricate software, hardware, or operational technology systems generate more complex problems and, thus, more detailed issue histories.
  • Maintain Compliance or Certification: Industries requiring high levels of compliance (e.g., ISO, SOC 2, medical devices) often necessitate meticulous documentation of all processes and changes.
  • Exhibit Operational Maturity: As outlined in who qualifies, companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with 50 or more full-time employees at peak (contractors excluded) and several years in operation are more likely to have accumulated a significant volume of historical data.
  • Possess Original Documentation: The data must be generated internally by the company, not sourced from third parties or public domains. It also must have clear rights to license.

Illustrative: A company in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE or Saudi Arabia that consistently logs every bug fix, every feature change, and every code review comment, including the discussions leading to a decision, possesses a highly valuable engineering issue history. This contrasts with a company that only documents final solutions without the context of the problem or the decision-making process.

What to Discuss (at a High Level)

When speaking with a potential referral, you can inquire about their practices around:

  • How they track and resolve technical issues.
  • How they document their engineering decisions and review processes.
  • The tools they use for internal knowledge management and project tracking.

Remember, you are looking for metadata-level information and indicators of robust internal processes, not the actual data itself. The company must always maintain full control over what data, if any, they choose to license with SourceX.

If you identify a company in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE or Saudi Arabia with these characteristics, encourage them to explore the potential value of their operational data. They can try our company fit checker for a preliminary assessment.

Rewards are calculated as the greater of $50,000 plus 5% and 15% of the compensation SourceX actually pays the company you refer, with no cap. A lead, meeting or signed agreement alone does not trigger payment. See the program terms.

What should a specialist advisor ask before referring a company for its engineering issue history?

Ask what the tracker cannot show from the outside: who wrote the issues and review threads, whose account holds them, whether any of them were ever licensed for AI training, and whether the CEO wants a conversation. Put the answers on a one-page engineering-history referral brief, filled in from what the engineering leader tells you and never from files. A long issue history is a reason to ask. Rights, fit with a current SourceX program and buyer interest are decided only in SourceX's review.

The indicators earlier on this page help you notice a company that documents its engineering work. What follows starts once you have noticed one: whose records they are, how to write them up honestly, and how to pass the idea to the person who can agree to it.

Two kinds of history inside every issue

An issue carries text that people wrote and a trail that the tool writes for itself. The text is the report, the comments, the design or code review discussion and any postmortem linked to it. The trail is the record of events: status changes, reassignments, label moves, and other issues or merge requests that point at it. GitLab's documentation calls these entries system notes. In substance, it says they show the history of events across the life of an issue or merge request, covering description edits, cross-references and changes to labels, assignees and milestones, and that an issue can be filtered to show comments only or history only. Other trackers present the trail in their own way, so ask rather than assume.

The split matters for a referral. The trail records the company's own process; the written text has authors, and authorship is where ownership questions begin. SourceX's supplier page on issue resolution histories says a company can license them "when it owns the records and removes customer and staff details first", and it names "Issues tied to clients whose contracts forbid sharing" among what is usually left out. Preparing records is a matter between the company and SourceX, so point an interested owner to that supplier guide instead of explaining it yourself.

The engineering-history referral brief

One page per company, written from a conversation with the engineering leader: the CTO, the VP of engineering or whoever runs the trackers. Beside each line, note the date and the name of whoever answered. The brief never needs an export, a screenshot, a sample issue or a login, and it goes nowhere until the CEO has approved its wording.

Brief lineWhat you recordWho can answerLeave out
CompanyLegal name, full-time headcount at its peak with contractors left out, and the main language of the recordsCEO or finance leadPayroll figures and staff names
Systems and yearsEach tracker, review tool and postmortem space, with the earliest year it still holdsEngineering leaderInstance addresses and credentials
AuthorsRough shares of issues opened by employees, contractors or agencies, customers, outside contributors and botsEngineering leaderAny issue text, even a title
Where the copy sitsThe company's own instance or organization, an agency's, a customer's, or one person's accountEngineering leader or IT administratorAny form of access
BreaksTool migrations, imported projects, deleted spaces, review threads lost on the wayEngineering leaderMigration files or logs
Earlier useAny past license of these records for AI training; any application, account, lead or email exchange with SourceXCEOThe contracts themselves
Machine-written textBot comments, AI-generated summaries, anything rewritten recentlyEngineering leaderExamples
PeopleThe internal sponsor, whether the CEO knows, and who will raise itSponsorOpinions about colleagues
Open for reviewRights for each author group, client contract limits, personal data, fit with a current programNobody yet: SourceX's reviewYour estimate of what the records are worth

Two lines carry more weight than they appear to. "Earlier use" catches the cases no reward can follow. SourceX's how-it-works page lists, among the points a supplier confirms, that its data "hasn't been licensed for AI training before, and it wasn't created or altered with AI for the sale". Section 3 of the terms adds that a company SourceX already knew through an application, an account, a lead or an exchange of emails counts as a duplicate, and no partner is credited for it. "Machine-written text" keeps routine bot output apart from people's writing, and it is the reason nobody should run old issues through an AI tool to tidy them before a call.

Three companies, three briefs (Illustrative)

Illustrative: the companies, systems and years are invented, and none of the outcomes predicts a SourceX decision.

Brief lineA: original workB: mixed third-party contentC: uncertain ownership
Company210-person maker of control software for packaging lines330-person developer-tools company with a public open-source project140-person freight-scheduling software company
Systems and yearsSelf-hosted tracker since 2013; reviews in merge requests; postmortems in a wikiPrivate tracker for the paid product; public issues since 2016; a service desk since 20202015 to 2018 in repositories under a co-founder's personal account; a company organization since 2018
AuthorsEmployees, plus build and dependency botsPrivate tracker: employees. Public project: mostly outside users. Service desk: customers writing by emailFounders and early staff, then employees
Where the copy sitsCompany serversCompany projects throughoutBefore 2018, an account the company does not control; since 2018, the company's organization
What the brief saysCompany original, with customer names in incident reports flaggedPrivate tracker described as the core; public and service-desk issues listed apartDescribes 2018 onward; marks the earlier years as unsettled
Left for SourceX's reviewClient contract limits and personal dataWhether any outside-written issue can be includedWho controls the early repositories and who owns the early text

A is the plain case, and still not a clearance. Employees wrote the text inside a system the company runs, which matches the documentation the company created itself that who qualifies describes. Incident reports still name customers and quote their configurations, so the brief flags them instead of calling the archive clean, and the bot comments are counted rather than hidden.

B shows how one tracker mixes authors. GitLab's Service Desk documentation, paraphrased, describes customers emailing bug reports, feature requests or feedback to a project address with no account of their own; each email becomes a ticket, which GitLab describes as an issue of a different type, and the customer goes on taking part by email alone. Add a public project where outside users open most issues under their own accounts, and the company's engineers have written triage, replies and fixes on top of text that other people wrote. The who-qualifies page counts "Material that belongs to the company's clients rather than the company" among what is usually not a fit. The brief does not try to settle the public or service-desk issues: it reports their shares and leads with the private tracker.

C is the case to slow down for. GitHub's documentation, paraphrased, says a repository owned by a personal account has one owner with full control, collaborators are the only other permission level, and passing ownership to someone else takes a repository transfer. The co-founder who holds the account left in 2019, and the repositories never moved. Engineers can still read the old issues as collaborators, which is access, not control. Two questions stay open: who controls the account, and whether everything written there, some of it before the company employed anyone, belongs to the company. The second is for the company's counsel and SourceX's rights review, not for the advisor. A careless brief would promise eleven years of history; the honest one describes eight and names three as held elsewhere.

Origin triage across other record families is worked through in separating owned content from customer-supplied material, and the code side of the same archive in private code and issue history.

Handing the brief to a sponsor, then to the CEO

The engineering leader is your sponsor; the CEO decides. Program rules give the CEO the authority to open a licensing discussion and to sign for the company. A CTO can describe the trackers and carry the idea upstairs but cannot start that discussion, and an introduction made behind the decision maker's back is usually not a fit.

  1. Ask to ask. "I introduce companies to a data licensing program. Could I ask how your team records issues and reviews? I won't ask to see any of it." A no ends the conversation.
  2. Ask bounded questions. Each one needs only a tool name, a year, a rough share or a yes or no:
    • Which tools hold issues, reviews and postmortems, and how far back does each go?
    • About what share of issues come from people outside the company?
    • Has any part of the history lived in an agency's, a customer's or one person's account?
    • Did a migration drop comments, review threads or the original authors?
    • Has any of it been licensed for AI training, and has the company ever applied to, held an account with or written to SourceX?
  3. Decline anything more. An offered sample, a screen share or a quick export is turned down politely; the records stay with the company.
  4. Let the sponsor correct the brief. The engineering leader reads the draft and fixes what you misheard.
  5. Let the sponsor choose who tells the CEO. The brief travels upstairs only with the sponsor's consent.
  6. Let the company apply. If the CEO wants the conversation, share your partner link so the company applies itself. Your SourceX Channel Partner Agreement must be signed before you submit anything (section 1), and credit goes to the earliest valid introduction inside the 90-day attribution window, once the company files a verified application (section 2).
  7. Step back for review. SourceX settles rights, scope, what gets removed and whether a current program wants the records. The advisor has no part in that review.

If a reward follows, SourceX pays it separately, and nothing is taken out of the company's agreed compensation to fund it (section 4). Whether to go through the engineering leader at all, or straight to the CEO, is weighed in introducing through a sponsor or directly to the CEO, and spotting the right internal person is covered in identifying an internal sponsor.

Where the brief can mislead

  • Shares are recollections. An engineering leader who joined in 2022 is estimating authorship for 2016. Mark old figures as estimates.
  • A label records authorship, not rights. "Company original" still leaves personal data and client contracts for review.
  • Imports can blur authors. If a migration changed who appears to have opened each issue, the authors line is weaker than it looks; ask, and write down the answer.
  • No mandate feedback is published. SourceX has not published which engineering histories its current programs want, so the brief leaves fit blank.
  • Method. Sources behind the brief, all read on 2026-10-27: terms version 2026-10-22.2, since replaced by version 2026-11-04, the who-qualifies page, SourceX's issue resolution histories and how-it-works pages, GitLab's documentation on system notes and Service Desk, and GitHub's documentation on personal-account repository permissions. It does not measure how often outside authorship or an uncontrolled account narrows what a company licenses. To measure it, keep each brief's author shares and compare them with the scope finally licensed; those figures are not yet published.

Next step

Refer a company. No cap on rewards. Default reward on a company's first qualifying transaction: the greater of $50,000 plus 5% and 15% of Actual Supplier Payout, with no cap; further qualifying transactions first agreed within 12 months of the first Closing pay 15%. Eligibility and payment conditions apply, including the USD 100,000 Minimum Transaction, and a signed Schedule A or addendum can differ. A lead, meeting or signature alone does not trigger payment.

When a brief shows history written by the company's own people, in accounts the company controls, and the CEO wants the conversation, register as a partner, check the company against who qualifies and read the program terms before anything is submitted.

Related: the metadata-only GitHub inventory template, what fractional CTOs notice during GitHub projects, the issue and sprint histories data page and the guides library.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Company signs and is paidThe company signs, delivers its data and is paid by SourceX under its agreement.
  4. Step 4You get your rewardthe greater of $50,000 plus 5% and 15% of the actual supplier payout, with no cap. A reward becomes payable after SourceX has paid the referred company under its signed agreement and the payout conditions in the program terms are met, including finance approval and your tax and payment details.

Common questions

What are engineering issue histories and review decisions?

This data refers to a company's internal records about problems in product development, software engineering, or operational technology. It includes issue tracking, troubleshooting, design reviews, post-mortems, and decision logs, often with metadata and code snippets.

Why is engineering issue history data valuable for AI developers?

AI developers use this data for pattern recognition, process improvement, and training AI that assists developers or automates debugging. It helps with risk mitigation and provides proprietary insights unique to each company's operations for AI labs and data buyers.

What types of companies have valuable engineering issue data?

companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia (outside the U.S., Canada and the EU, at least 70% of the data must be in English) with 50+ employees at peak, using robust internal tools like Jira, practicing structured development methodologies (e.g., Agile), and having a culture of documentation often possess valuable data. They should also operate complex technical products and maintain compliance.

What should I discuss with a potential referral about their data?

You can inquire about how they track and resolve technical issues, document engineering decisions and reviews, and the tools they use for internal knowledge management. Focus on indicators of robust internal processes, not the data itself.

May I tell SourceX about a company's issue tracker before its CEO has agreed?

No. The brief stays between you and the engineering leader until the CEO has approved its wording. SourceX's screening guidance treats an introduction the decision maker did not know about as usually not a fit, so an early mention helps nobody.

Do bot comments and AI-written summaries in a tracker count against the company?

The brief lists them as machine-written so the review can decide what to include. What the company must avoid is producing or rewriting records for the deal: its supplier confirmation says the data 'wasn't created or altered with AI for the sale'.

The engineering leader offered me one resolved issue to show the quality. Should I accept it?

Decline it. A sample is something the company prepares for SourceX after it applies: SourceX's page on issue resolution histories describes a few dozen resolved issues from one project, cleaned and approved by the company. A partner never needs to see one.

Sources

  1. docs.github.com/en/repositories/managing-your-repositorys-se…
  2. docs.gitlab.com/user/project/service_desk
  3. docs.gitlab.com/user/project/system_notes
  4. refer.sourcex.si/terms
  5. refer.sourcex.si/who-qualifies
  6. sourcex.si/data/issue-histories
  7. sourcex.si/how-it-works

Free resources

Know a company with valuable proprietary data?

Become a referral partner, get your link and introduce the company’s CEO (in an owner-run company, usually the owner). Referral partners can join from most countries. We can't work with partners in countries or regions under U.S. sanctions or on U.S. restricted-party lists. The companies you introduce should be based in the U.S., Canada and the EU, or in the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia with at least 70% of their data in English.

Refer a company →

I own a business

Explore licensing your company's data directly to SourceX for AI training and evaluation. Start a short assessment; no uploads needed.

Start an assessment