Search guides, questions and resources

Recordings made on a company's Zoom or Microsoft Teams account are generally company records, not the host's personal property, while the vendor stores and processes them under its customer agreement. Ownership does not settle use: vendor terms, the company's recording policy, notice to participants and outside parties' rights decide what can be done with them.

It depends on the model, but the payer is almost always the party that receives the new business, not the person referred. At SourceX, the partner reward comes out of SourceX's own collected fee. It is never deducted from what the referred company receives, and it is paid only after the buyer pays.

The portfolio company receives the proceeds. As licensor, it is paid one all-in, one-time price, with SourceX's fee included, typically within about 60 days of invoicing once the buyer selects the data. Whether any of that cash later moves to the fund is a separate decision for the company's board, subject to its governing documents and lenders.

Few people see a company's data early. The partner sees basic fit information only, SourceX sees what the company shares during qualification and inventory, buyers review a described dataset before signing, and delivery happens only after an executed agreement and the company's authorization.

AI training data is sold or licensed by five kinds of supplier: publishers, online platforms, data vendors, human-data firms that produce examples to order, and operating companies licensing their own business records. The last group is the least organized, so for US companies with 50+ full-time employees at peak, an introduction is often where the path to buyers starts.

The cleanest split gives the operating team ownership of a value creation initiative like data licensing and leaves the board relationship and exit story with the deal team. The operating partner screens companies and introduces the CEO, the deal partner approves timing against the exit plan, and one named person registers as the SourceX partner and tracks each referral.

A private equity sponsor should set one routing rule before its first introduction: reward paid to the individual, to the management company, or offset to the fund. The right choice depends on the fund agreement, conflict policy and investor expectations, so decide with counsel and your compliance officer.

Tell the people who must decide first: the owners, the board where there is one, and the authorized signer. Bring in counsel and IT next, HR before scope is final, then managers, and staff once scope and exclusions are clear. A short RACI table keeps the order from drifting.

Wholesale distribution M&A in 2026 is shaped by consolidation from public distributors and private equity platforms, a wave of family-owner transitions and tariff-driven caution about margins. Owners timing a sale can strengthen their position by organizing ERP quoting, pricing and order histories, which buyers diligence closely and which AI developers may license for a one-time payment.

A functioning licensing market matters because fair use analysis asks how a use affects the potential market for the original work. Where buyers can license training data, unlicensed copying looks more like a substitute for a sale. It is one factor, not a verdict, which is why permissioned deals are attractive to AI developers.

AI agent builders want CRM and ERP history because real approvals, exceptions, configuration changes and deal outcomes show how work actually happens in business software. ERP, CRM and RevOps consultants can spot clients with licensable histories and introduce them to SourceX without exporting or handling any client data.

AI agents fail at real business tasks mainly because small errors compound over many steps, they lack the unwritten context staff rely on, they misuse tools and permissions, and they rarely recover once off track. Behind most of these is a data gap: models have seen finished documents, not records of how people actually completed multi-step work.

AI labs and data buyers want records from defunct and closing companies because years of real work, such as tickets, code reviews, workflows and decisions with known outcomes, help train and evaluate AI agents and are scarce on the public web. A closure often destroys those records, so the window to license them is before systems are cancelled.

AI developers want US company data because it holds English-language workflows, US business conventions and one legal framework for rights review. Public text is finite, and agents need records of real work. That is why introductions must be US companies, while partners can join from any supported country.

AI labs value records of mistakes because they preserve the failed attempt, the diagnosis and the fix, which clean final documents hide. Reopened tickets, postmortems, credit memos and rejected pull requests are examples. Partners can treat a messy archive as an asset when the original, the correction and the reason all survive.

AI pilots at portfolio companies usually fail for data and organizational reasons rather than model quality: records scattered across many systems, no outcome labels, weak integration with core tools, unclear permissions and no process owner. The discovery work a stalled pilot leaves behind, especially its map of systems, history and access, is reusable groundwork for a data licensing review.

AI data buyers are often kept confidential because their data sources are competitively sensitive and agreements commonly restrict disclosing the deal. Owners are protected not by knowing the name on day one but by seeing price, term, purpose and obligations before signing. Partners should never name or guess a buyer.

AI licensing deal terms are usually confidential because the contracts commonly carry confidentiality clauses that protect both the data owner's relationships and the buyer's strategy. What the public sees is a thin slice: announcements without numbers, aggregate filings or unconfirmed leaks, none of which prices a private company's records.

Linked records across systems are worth more than single exports because they show whole workflows: the request, the decision, the work and the outcome. AI buyers training agents need that sequence. For partners, a company running 10-15+ connected systems is a positive signal, not a complication.

A US payer typically asks for a Form W-9 so it has your correct taxpayer identification number to report payments it makes to you to the IRS. You give it to the payer rather than filing it with the IRS, and it is different from a 1099, which the payer may issue later.

AI models need large and varied data because they learn patterns from examples; pretraining uses huge public corpora, while post-training and evaluation need records of real work. As public text is reused, new kinds of records, such as business workflows and outcomes, become more valuable than more of the same.

Metadata raises the value of business data because it records when, by whom and with what outcome each item was created, turning documents into sequences AI developers can learn from. Partners should ask one question before introducing a company: will exports keep dates, authors and links between records?

Pre-AI data has value because records a company created before generative AI tools went mainstream at the end of 2022 are reliably human-made, with real decisions and outcomes attached. As more new text is machine-assisted, AI developers prize verifiable human records, so a company's 2010-2022 email, tickets and project files are worth reviewing before they are lost.

Realistic RL environments for business software depend on real company workflows because ticket queues, approvals, CRM stages and month-end close supply the starting states, action sequences and recorded outcomes that tasks and graders need. Invented scenarios miss the exceptions and handoffs that real operations records contain.

Not automatically. IRS Publication 515 describes 30% as the general US tax rate on most US-source income paid to foreign persons, unless a treaty reduces it. Whether that reaches a referral reward depends mainly on whether the reward is US-source, which for services turns largely on where the work was done, and on the W-8 form the payer holds.

Licensing records to AI developers is unlikely, on its own, to replace your business, because models learn general skills from many sources and a license covers only the records you agree to include. The real risk is competitive, so the license scope, exclusions and term matter. Nothing is binding until you sign.

A data license does not have to get in the way of selling a business, but a buyer will find it in diligence and read its terms. A documented, time-limited license over records the company still owns is generally manageable. Trouble comes from undisclosed licenses, unclear rights, broad exclusivity or a license signed mid-process without the buyer's consent.

It can if customers feel their information was handed over, and usually does not when the license covers the company's own de-identified workflow records and the owner is ready to explain it. Reputation risk depends on whose data is included, how it is described, and whether counsel agreed a disclosure approach.

Not if you control three risks: pressure, undisclosed fees and overpromising. Ask permission first, disclose any reward in writing, and make clear that nothing is binding until the company agrees price and terms and signs. Most damage comes from how the introduction is made, not from the introduction itself.

A data licensing price is agreed between the company and SourceX before anything is binding, as one all-in price with SourceX's fee included. Partners never quote or estimate it. What moves value is breadth, history, outcomes and rights clarity, and the company decides whether to sign.

A company is not named by default when it licenses data to an AI developer. Publicity is a contract term the owner negotiates before signing, and deal terms are usually confidential. Disclosure rules for AI developers vary and change, so owners should confirm current requirements with counsel.

Synthetic data is unlikely to replace real data for most business uses. Generators need real seed examples to imitate, and their output has to be checked against real outcomes, so authentic, rights-cleared company records stay scarce. Synthetic data does reduce demand for generic, easily imitated material, which is why depth, outcomes and clean rights matter more than volume.

A wind-down budget should include a line for keeping key systems alive, or for verified exports into controlled storage, until the records have been assessed. Early subscription cuts destroy the longest histories, and the line costs money now to protect an option that disappears with the archive.

A quick-turn archive sale suits small venture-backed startups closing fast; a managed licensing process suits established companies with 50+ full-time employees at peak, years of records and an owner who wants control of scope, price and terms. SourceX runs the managed route, from qualification and rights review to delivery and payment.

A wind-down engagement letter should name records preservation, who controls system access, and any data licensing review as in-scope or out-of-scope work, and should disclose a referral relationship with SourceX if one exists. The clauses below are adaptable starting language for counsel to review, not a finished contract.

A workflow map template lists each step of a business process, the system that holds its records, years of history, who can export and who owns the rights, using metadata only. It shows AI data buyers the breadth of connected records across systems that a plain list of tools misses.

Accounting firms can add one optional paragraph to the year-end client email inviting owners with long operating histories to ask about licensing business records. Include a plain disclosure line, check AICPA and state board rules first, and send it only to clients where the rules allow.

A year-end close checklist should cover cut-off, reconciliations, adjustments and workpapers, and can add a metadata-only records note listing systems and renewal dates without opening any data. CPAs should check the AICPA commissions and referral fees rule and their state board before making or accepting any referral.

A year-end exit planning review should cover entity documents, key-person cover, tax and structure, and the company's systems and archives before January renewals or retention settings erase history. Advisors can run it in one Q4 meeting using the checklist below, without handling any company records.

A year-end tax planning meeting with a business owner should cover income and deduction timing, owner pay and distributions, estimated payments, next year's purchases and system changes, and big decisions ahead. For owners of US companies with 50+ full-time employees at peak (contractors excluded), add one question about legacy archives: it can surface a permissioned data licensing introduction.

A business network referral shortlist worksheet helps SourceX partners systematically evaluate potential US company referrals against key qualification criteria. It guides partners to identify companies with 50+ employees, appropriate data rights, and an authorized sponsor, without collecting sensitive company data.

Fractional CTOs can effectively integrate data licensing referrals into their practice by dedicating the first month to identifying and qualifying client companies against SourceX's baseline, securing an authorized sponsor, and making the initial introduction. This structured approach helps maximize referral potential without disrupting ongoing client work.

A successful first month for MSPs in data licensing referrals involves identifying eligible US-based clients, educating them on the opportunity, and making a warm introduction to SourceX. Focus on clients with 50+ employees and clear data licensing rights.

A first-month data referral plan for venture platform teams should focus on identifying suitable portfolio companies, preparing them for an introduction, and making the initial connection with SourceX. It's crucial to understand the referral criteria and the attribution process to maximize success.

For business brokers, a successful first month in data licensing referrals involves identifying suitable clients, strategizing effective introductions, and diligently tracking the attribution window. This guide outlines a practical, step-by-step approach to get started efficiently.

A good first month referral plan for CX and contact center consultants involves screening existing clients against SourceX's company baseline, identifying an authorized sponsor, and making a timely introduction. Focus on understanding the program mechanics and setting up a clear tracking system.

A first-month data referral plan for ERP consultants should focus on identifying suitable US companies with 50+ employees and rights to license data, making initial introductions, and diligently tracking the attribution window. This structured approach helps maximize referral potential within the program guidelines.

A good first month referral plan for software implementation partners involves systematically identifying existing US clients that meet SourceX's baseline criteria and have an authorized sponsor, then making warm introductions. Focus on establishing a consistent process for client screening and initial engagement.