This hub offers practical guides and best practices for advisors looking to successfully introduce 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) to SourceX, a direct licensing partner for approved business data for AI training and evaluation. Here, you'll discover step-by-step instructions and strategic insights for introducing companies to SourceX, which licenses data directly from companies and manages the agreement, rights review, data preparation, delivery and payout, subject to acceptance at Closing and agreed terms. These resources are designed to help you navigate the process efficiently, focusing on companies with 50+ full-time employees at peak. Unlock the potential of valuable data partnerships by leveraging these comprehensive guides.
Practical guides to successful company data introductions
Short, task-focused guides on identifying suitable 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), recognizing valuable proprietary data, and making an introduction the decision maker will welcome.
- Step 1Share your linkSend your personal link to a company you know.
- Step 2Company appliesThe company applies itself at /apply.
- Step 3Company signs and is paidThe company signs, delivers its data and is paid by SourceX under its agreement.
- 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 kind of companies should I introduce to SourceX?
You should focus on introducing 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 have 50 or more full-time employees at peak. These are the companies that can unlock the potential for valuable data partnerships with AI labs and data buyers.
What is the purpose of these guides?
These guides offer practical instructions and strategic insights for advisors to successfully introduce companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia to SourceX. They help navigate the process efficiently and identify valuable proprietary data for AI labs and data buyers.
Who benefits from SourceX data introductions?
AI labs and data buyers benefit from approved business data that SourceX licenses directly from companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia for AI training and evaluation. Advisors can also benefit by introducing companies to SourceX as their direct licensing partner.
What kind of information do these guides provide?
These guides provide step-by-step instructions and best practices for identifying suitable companies in the U.S., Canada, the EU, the UK, Australia, New Zealand, India, Singapore, South Africa, Latin America, the UAE and Saudi Arabia. They help in recognizing valuable proprietary data and making introductions that decision makers will welcome, facilitating data transactions for AI.
Related pages
- How to model a one-time receipt, like a data license, in a 13-week cash flow forecast
- 2021-vintage private equity funds: what to do with companies bought at peak multiples
- 3PL and logistics M&A in 2026: what buyers want and what sellers should preserve
- 3PL valuation multiples in 2026: what buyers pay for in a logistics business
- A/E industry M&A trends in 2026, and what selling firms should prepare
- Acquiring a business from a retiring owner: assessing decades of records before they go
- Advisors: Unlocking Data Value for Transportation Clients
Transportation advisors can introduce their clients to SourceX for data licensing, helping them monetize operational data. The role involves identifying qualifying companies, raising the opportunity with an authorized sponsor, and understanding the reward structure.
Read → - After an acquisition, the identity map decides whether an archive still joins up
Identity changes after an acquisition can split one person, project or customer into two records that no longer join. Renamed user accounts, re-keyed projects, merged domains and new legal entities each break links between pre- and post-deal data unless someone kept a crosswalk mapping old identifiers to new ones. An archive whose joins survive tells a continuous story; one whose joins break reads as two unrelated datasets. Ask whether the crosswalk exists before a licensing conversation, not after.
Read → - Agency Owners: Your First Month Referral Plan for Data Licensing Referrals
Agency owners can kickstart data licensing referrals by identifying clients who meet specific criteria, securing an authorized sponsor, and making a timely introduction. This guide outlines a structured approach for the first month to set you up for success.
Read → - Agent recovery data: a failed chat transfer, the fix, and the records that prove it
Agent recovery data is the record of a task that went wrong partway through and was then brought back to a verified, correct result. In an ordinary support workflow it looks like this: a chat transfer to a human that never completed, the ticket that picked the case up the next morning, the refund that fixed the problem, and the quiet week afterwards that shows the fix held. The failure, the repair and the proof all have to be linked to one case.
Read → - Agentic AI and PE-backed services companies: the risk, and the records hedge
Agentic AI puts pressure on PE-backed services companies whose margins depend on people doing repeatable, documented work such as support, IT operations, claims handling and back-office processing. The ticket histories, SOPs and QA records that make that work automatable are also what AI developers license, so a qualifying company can earn a one-time payment while it adapts.
Read → - Agentic AI in accounting in 2026: what it means for CAS firms and their clients
Agentic AI in accounting in 2026 means software agents that carry out multi-step finance work, such as coding invoices, matching payments and running close checklists, with people reviewing the results. Building those agents takes real finance workflow records, so CAS firms whose clients hold years of AP, close and reconciliation history may have a permissioned introduction to make.
Read → - Agentic AI in private equity portfolios: deploy agents, and value the records behind them
Agentic AI gives a private equity portfolio two levers. Portfolio companies can deploy AI agents to take cost out of support, finance and operations, and the same companies hold years of records of real multi-step work that AI labs need to build those agents. Through SourceX, rights-cleared historical records can be licensed on agreed terms.
Read → - Agentic AI in the enterprise in 2026: why agents need records of real workflows
In 2026, enterprise agentic AI means software agents that carry out multi-step work across business systems, such as triaging tickets, matching invoices or updating CRM records, rather than only answering questions. A growing constraint is the examples, not the model: agents learn from records of how real workflows run, fail and finish, which only companies hold.
Read → - Aggregated and de-identified data clauses in B2B SaaS contracts: what they allow
An aggregated data clause lets a vendor reuse customer-derived data after it is aggregated or de-identified, usually for stated purposes like product improvement. It does not automatically allow licensing to third parties for AI training. Purpose, definitions, confidentiality terms and public promises decide the answer, so counsel should read the clause first.
Read → - AI agent benchmarks explained: why labs want private business tasks
AI agent benchmarks are standardized task sets that score whether a model can finish multi-step work. Public ones saturate and leak into training data, so AI labs increasingly want private, real-world task records from companies. That evaluation demand is one reason operational records with recorded outcomes are valuable to license.
Read → - AI agents for ERP: why process histories matter to model builders
AI agents for ERP are software that carries out multi-step finance and supply-chain work, such as matching invoices or routing approvals. Training and evaluating them needs records of real approvals, exceptions and outcomes, which sit inside companies, not on the public web. That makes licensed ERP histories scarce and valuable.
Read → - AI agents in IT service management: what training data do they need?
AI agents for IT service management need records of real work: tickets linked to their resolutions, change approvals, escalations and runbook steps. Those histories sit inside MSPs and IT-heavy companies, not on the public web, so licensed, rights-cleared ITSM records are scarce and valuable to AI labs and data buyers.
Read → - AI data licensing trends for 2026 and what they mean for referral partners
The main AI data licensing trends for 2026 are a shrinking supply of fresh public text, rising demand for records of real work that agents can learn from and be tested on, tighter privacy rules, and private equity's turn to operational value creation. Together they favor established companies with rights-cleared, well-documented business records.
Read → - AI in M&A in 2026: what to tell sell-side clients before launch
Sell-side advisors should tell clients that buyers increasingly probe AI exposure and that proprietary records may be licensable separately. Prepare a one-page AI narrative and a system-by-system records inventory before launch; SourceX can then handle licensing of qualifying records outside the sale itself.
Read → - AI in private equity in 2026: where value shows up and the lever most plans miss
AI in private equity in 2026 is mainly a cost and productivity program inside portfolio companies, with results that vary by company and depend on data readiness. Most plans miss one lever: licensing a portfolio company's historical operating records to SourceX, which depends on rights and history rather than tool adoption.
Read → - AI notetaker consent laws: when can you record and transcribe a meeting?
Whether an AI notetaker needs everyone's consent depends on where participants are and whether the conversation is confidential. Federal law allows recording when one party consents, but California and some other states require all parties' consent for confidential communications. Announcing the notetaker and keeping consent records protects the company and decides whether transcripts can ever be licensed.
Read → - AI representations and warranties in M&A: what buyers ask and how sellers answer
AI representations and warranties in M&A purchase agreements cover the seller's use of AI tools, its rights to data used to train models, compliance with privacy and contract promises, and any licenses letting others train on company data. Sellers answer through the disclosure schedule; one holding an executed, documented data license has a clean exception to list.
Read → - AI roll-up integration: how to preserve pre-automation work records before rebuilding
An AI roll-up integration playbook should start by inventorying and archiving each add-on's human-only work history, such as tickets, files, approvals and decision trails, before AI tooling changes how work is recorded. Keep client-owned material separate, then screen each add-on with 50+ full-time employees at peak for a possible SourceX data license.
Read → - AI roll-up strategy explained: what services-firm owners and their advisors should know
An AI roll-up strategy buys established services firms, rebuilds delivery around AI tools and agents, and consolidates them on one platform to raise margins. Because these acquirers prize workflow records, owners and their M&A advisors should map which records the firm owns, which belong to clients, and whether they have independent licensing value before signing an LOI.
Read → - AI strategy consulting engagement deliverables, plus the data asset review most skip
Typical AI strategy consulting engagement deliverables are a current-state assessment, a ranked use-case portfolio, a data and technology readiness review, a governance outline and a phased roadmap. Add one short workstream that asks the reverse question, which records AI developers might license, and end it with an owner decision on requesting a SourceX introduction.
Read → - AI task length keeps doubling: what METR's research means for business data
METR, an independent AI evaluation nonprofit, measures an agent's time horizon: the length of task, timed by how long skilled people take, that the agent completes about half the time. Its 2025 research reported that this horizon has been doubling at a steady pace for several years. Longer autonomous work needs training records of complete, multi-step tasks.
Read → - AI value creation in private equity: a playbook for operating partners
AI value creation in private equity runs on two tracks. The buy side applies automation, copilots and agents to cut costs or grow revenue, and depends on clean data and scarce talent. The supply side licenses a portfolio company's years of work records to SourceX, and needs no AI build or new hires.
Read → - AI washing and the SEC: how PE firms should describe portfolio AI and data initiatives
AI washing is overstating how much a firm or its portfolio companies actually use artificial intelligence, and the SEC has pursued investment advisers over misleading AI claims. A PE firm should keep every AI or data claim in fund marketing, DDQs and LP letters accurate, substantiated and labeled by status, including data licenses that are not yet signed.
Read → - AI washing in due diligence: how to test a target's AI and proprietary data claims
AI washing due diligence checks whether a target's claims about its AI and its proprietary data survive evidence. Test capability claims with architecture and vendor records, and test data claims on three proofs: the company has rights to the records, the history is as deep as stated, and someone can export it today. Unproven claims belong in the price.
Read → - AI-generated code in due diligence: the questions buyers ask and how sellers answer
In due diligence on AI-generated code, buyers ask which AI coding tools engineers used, under what terms, where AI-assisted code sits, how it was reviewed, and whether ownership and open-source obligations are clean. Sellers answer with a written tool policy, plan and settings records, and the pull-request review trail. That human-authored engineering history is also what AI developers license.
Read → - AI-native ERP: what it is and what happens to your legacy history when you switch
An AI-native ERP is finance and operations software built around automation from the start, with AI coding transactions, matching payments and preparing reconciliations, rather than AI features added to an older system. Switching usually moves balances and recent detail only, so CFOs must decide where older ERP history lives and whether it holds value.
Read → - AI's impact on SaaS valuations in 2026, and the options software owners have
AI pressures SaaS valuations in 2026 through the questions acquirers ask: whether seat-based revenue survives AI agents, whether AI-native rivals can copy the product, and whether AI features are real. Owners answer with retention and product evidence. Licensing engineering, support and product records directly to SourceX is separate: one-time cash outside the multiple, not a valuation fix.
Read → - AICPA confidential client information rule: what a CPA can share in an introduction
The AICPA confidential client information rule (ET 1.700.001) bars a CPA in public practice from disclosing confidential client information without the client's specific consent. Before introducing a client to SourceX, get that consent, share only the company name, a sponsor contact and owner-confirmed fit facts, and check ET 1.520 and your state board's referral-fee rules.
Read → - AICPA conflicts of interest: what a CPA must do when an introduction pays
A paid introduction is a self-interest threat under the AICPA conflicts of interest interpretation, so a CPA must identify and evaluate it, then apply safeguards, disclose and obtain client consent, or decline. Keep the owner in charge of the decision, introduce only companies that fit the published criteria, and document everything in the client file.
Read → - Ajera to Vantagepoint migration: why to keep the full project history
In an Ajera to Vantagepoint migration, archive the full project, time and billing history even if only recent years convert. The archive preserves records that may later support a data licensing review, and a Deltek consultant can introduce a qualifying A/E firm with 50+ full-time employees at peak to SourceX without touching the data.
Read → - Alert deduplication decisions: what to inventory before a monitoring tool replacement
Before a monitoring tool is replaced, inventory four things: the grouping and suppression rules in force over time, the alert groups each rule produced, the analyst notes that merged or split those groups, and the final incidents they fed. Record for each where it lives, how many years it covers, who owns it and when access ends. That history explains how an operations center told one outage apart from unrelated noise, and it is the part migrations often leave behind.
Read → - Alert deduplication system map: the keys that join alerts, incidents, notes and rule changes
Five joins rebuild one alert deduplication decision across systems: each alert to its group through a dedup key or label set, each group to its incident through an incident ID, each incident to the analyst's reasoning through its log entries, each incident to the configuration edit it prompted through a ticket or commit that cites it, and every timestamp to one normalized clock. Any join fails when an export drops the field it depends on.
Read → - Alert storm worked case: how one storage fault became three incidents, then one
The records that explain an alert-handling exception are the ones made where automation got it wrong and a person corrected it: the alert groups as the tooling formed them, the analyst's merge with a written reason, the rule change that followed and the next incident that proved the change worked. In the Illustrative storm below, one storage fault produced 147 alerts and three incidents before an analyst tied them together.
Read → - Alert-to-repair-verification record linkage: the keys that join each step, and the link that usually breaks
By carrying one key forward at every step: the asset ID from the sensor alert to the work order, the work order number to the visit, the parts issue and the status change, and the sensor's own identity into the readings that verify the repair. Most companies hold the first keys in system fields. The link that usually breaks is the last one, because a sensor swapped during the repair often keeps the old data stream, so nothing records which instrument produced the verification readings.
Read → - Alert-to-repair-verification records across IoT, Salesforce Field Service and ERP: what proves a fix worked
A referral partner can describe a company's alert-to-repair-verification records from metadata: which IoT platform holds sensor readings and alerts, how work orders and visits are kept in Salesforce Field Service, which ERP records parts and invoices, the years and asset counts each covers, the identifier that ties a reading to a work order, and who owns each system. The case to test is a sensor replaced during a repair, because the new sensor resets what "normal" means.
Read → - Alert-to-runbook-revision records: tracing Datadog, PagerDuty and Confluence history when the alert clears before the fault does
A company connects alert-to-runbook-revision records by tying each Datadog monitor alert to the PagerDuty incident it opened, the responder's notes and postmortem, and the Confluence runbook version that changed because of it. The links are monitor IDs, incident numbers, timestamps and page version comments. A referral partner can describe which links exist, for which years, and who owns them, without opening an alert or a runbook. The trail usually breaks when an alert clears before the underlying fault is fixed.
Read → - An archive custodian is not always the licensing decision-maker: a custodian-versus-authority checklist for whoever holds the copy, built on the confirmations a licensing company has to give
Holding an archive gives you knowledge of it, not authority over it. A custodian becomes the licensing decision-maker only when it is also the party SourceX's screening page calls the company's controller: the company itself acting through its CEO, an acquirer that took the records, or a trustee, receiver or assignee controlling the estate in a formal process. Everyone else with a copy, from a storage vendor to an IT firm keeping backups to a former administrator, answers questions about the archive's condition and passes the licensing question to whoever it holds the archive for.
Read → - An archive introduction should start with the current custodian: six status questions for whoever holds the copy, and a custody-versus-authority table that turns the answers into ask, wait or stop
Start with whoever holds the archive today, because only the holder can tell you whether the records still exist, which deletion or hold clock is running and whose instructions it follows. That holder is a witness, not a decision-maker. Follow its answer back to the entity it takes instructions from, and ask that entity's CEO whether the company wants a licensing conversation. A clear holder with an unclear instructing party means wait; no confirmed holder means stop.
Read → - An eligibility review that ends with a clear yes, no or later: a three-outcome rubric for every company your fund could introduce
Every company your team screens should leave the review with one verdict. **Yes**: introduce now, because the published fit tests pass on dated facts and the CEO wants the conversation. **No**: a test fails in a way time will not fix, such as records that belong to clients. **Later**: a test fails or is still unknown, and a named trigger and date reopen it. None of the three is an acceptance; SourceX decides fit in its own review. First settle which legal entity holds the records and who controls it today.
Read → - An EOS IDS example: should we license our operating records?
To run an EOS IDS on licensing company records, identify the real issue (usually rights, sponsor or archives), discuss the facts for about ten minutes, and solve to one owned next step such as a fit check, not a decision to license. The worked example below is fictional and shows the mechanics.
Read → - An EU portfolio company with English records: questions before introduction
Before referring an EU portfolio company, an introducer should ask five screening questions: how many full-time staff the specific entity had at peak, how long its records go back, which systems hold them, who owns them, and who can authorize a licensing discussion. Those questions decide eligibility. Questions about GDPR lawful basis, purpose compatibility, client contracts and AI Act transparency belong to a separate rights review that SourceX runs with the company and its counsel after the company applies.
Read → - An evidence-based scorecard for choosing an AI data referral program
A useful scorecard records what each program's written terms say on twelve fields: who pays the reward, the calculation base, the rate or amount, the cap, the cap's scope, the payment trigger, the deadline, reversals, discretion, attribution, company eligibility and the terms version. Score each field on disclosure, not size. Leave a field marked "not published" rather than guessing, and never treat a blank as zero.
Read → - An investment team's reading list can become a platform agenda: a research-to-action template that admits each paper only beside the paper that argues against it, gives it one narrow network test with a stop rule and retires it on a set date
A VC research platform agenda turns what the investment team is reading into a short list of questions the platform team can test with CEOs it already knows, not a list of companies it endorses. Admit a paper only beside a credible paper that argues against it, give it one narrow test that needs no company data and carries a stop rule, name an owner, and retire the item on a set date. Results stay descriptive, and every CEO who declines or never answers stays in the count.
Read → - An operating partner's map of data owners and gatekeepers: draw who owns each record, who can ask the CEO and who can stop the conversation
In most portfolio companies the records have several owners and the licensing decision has one: the CEO. Data owners are the executives accountable for each record set, such as the CFO for finance systems or the head of support for tickets. Sponsors can raise the topic with the CEO and be heard. Gatekeepers, such as general counsel, a data protection officer or IT security, can slow or stop it. Map all four roles per company before you raise data licensing, and route the ask through a sponsor to the CEO.
Read → - An unpaid introduction can still require client consent: a direction-of-information checklist that asks what would leave your file, to whom and on whose authority, before anyone raises the question of pay
Declining a reward changes what you must disclose about money. It does not change what you may reveal about a client. The confidentiality rules lawyers and CPAs work under are written around client information, not compensation, so telling SourceX a client's name, size or records can raise a consent question even when you take nothing. The route that reveals nothing is telling the client's CEO that SourceX exists and letting the company apply itself.
Read → - Annual license true-up for SaaS spend consultants: will the proposed cancellation delete unique decision histories?
Sometimes, and the people building the true-up cancel list rarely ask. A cancellation deletes unique decision histories when the app is the only place where years of approvals, case resolutions, deal notes or design decisions were recorded, and when no export happens before the vendor's retention period ends. SaaS spend consultants can test each app on the cancel list with one question. If the company holds such records, looks eligible and wants to explore licensing, the consultant introduces its decision-maker to SourceX.
Read → - Antitrust basics for associations discussing data with members
A trade association can host neutral education about data licensing, but it should not collect, pool, compare or circulate members' records, prices or terms. Each member should license its own data independently. Antitrust analysis is fact-specific, so association counsel should review agendas, surveys and materials first.
Read → - Appeal to course correction: linking a rubric dispute, its decision and the curriculum fix across LMS, assessment and support records
A referral partner can describe a training company's appeal-to-course-correction history without opening a learner's file. The facts that matter are metadata: which system receives the dispute, where the appeal decision is recorded, how rubric and course versions are kept in the LMS, how many appeals lead to a change each year, which identifier ties them together, and who owns each step. The case to test first is a rubric fix that was applied to a different course version from the one the appeal was about.
Read →
Free resources
- Business succession planning assessment — Ten questions on successor, transition and documentation.
- NPV calculator — Net present value with a discounted cash flow table.
- Time value of money calculator — Future and present value with optional regular payments.
- All free tools · MCP resource center
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