Short answer
Some AI developers, and the firms that supply them, pay for copies of real company records that show how work gets done step by step. The developers use them to train AI agents. A logistics business may be able to license a de-identified copy of its records for AI training without selling the business itself. What can be included depends on who controls each record, what your contracts say and which laws apply, and no one can promise a price before a review.

Why AI developers want company data for training
AI developers have run short of new public data. Forbes reported in April 2026 that labs had exhausted the public internet by late 2024, and that public web text is less useful for building "agentic" AI that does actual work. MIT Technology Review reported in October 2026 that demand for company archives "emerged suddenly" in 2025.
What buyers describe wanting is real work, in sequence. One buyer told MIT Technology Review it looks for archives with disagreements and difficult judgment calls, and uses them to build simulated workplaces where AI agents practice tasks. Forbes reported around 50 startups building simulated workplaces of this kind. As Fast Company summarized, AI agents need documents, emails and FAQs that carry context and feedback.
In a 2024 benchmark published by Sierra, which tested AI agents on tasks such as rebooking airline flights under policy rules, the best model completed fewer than half of the tasks on average.
One caution: no company has published tests showing how much real company data improves AI models, according to MIT Technology Review.
Which logistics data AI developers may want
One firm told Forbes it values linked records, such as a work ticket tied to the code change it led to, over standalone documents. In logistics, that might be a quote, the load it became and the claim that followed. What each type of business holds, and what usually limits sharing, differs a lot:
| Business type | Records you may hold | What usually limits sharing |
|---|---|---|
| Freight brokers and agents | Quotes and awards, rate confirmations, carrier paperwork, emails, claims | Shipper confidentiality (in the US, 49 U.S.C. 14908), NDAs, load-board terms |
| Freight forwarders and NVOCCs | Shipment files, bookings, bills of lading, quotes, agent correspondence | In the US, 46 U.S.C. 41103; carrier contract rates |
| Customs brokers | Entries, classification and valuation work, client correspondence | In the US, the client's written authorization (19 CFR 111.24) |
| Trucking and haulage | Dispatch, settlements, maintenance, claims | Shipper confidentiality, driver testing records, telematics terms |
| LTL and regional carriers | Freight bills, weighing and inspection records, reclass outcomes, claims | Shipper pricing and confidentiality |
| Drayage and intermodal | Dispatch, appointments, gate times, interchange receipts, disputes | Shipper confidentiality (in the US, 49 U.S.C. 14908) |
| Cold chain logistics | Temperature logs, setpoints, alarms, written procedures | Shipper confidentiality (in the US, 49 U.S.C. 14908) |
| Couriers and last-mile delivery | Proof-of-delivery photos, signatures and scans, GPS data, recipient details, courier metrics | Recipients' and couriers' personal data |
| 3PL and contract logistics | Contracts and rate cards, billable activity records, EDI messages, exception records | Customer data is generally treated as the customer's |
| Warehousing and fulfillment | Inventory records, cycle counts, scans, billing events | Merchants' data, workers' speed data, WMS terms |
| Shipbrokers and chartering | Chartering emails, fixture recaps, charter parties, laytime calculations | Principals' confidential terms, exchange data |
| Ship agents and port agencies | Port call files, disbursement accounts, correspondence | Under the 2001 FONASBA agency form, files belong to the principal |
| Ship owners, operators and managers | Noon reports, maintenance histories, safety management records, P&I claims files | Under SHIPMAN 2024, vessel data is the owner's; crew data |
| Closed or winding down | Whatever the business held while it was operating | Who can decide once a formal process starts; software deletion |
Each sector page sets out which records are usually the business's own and which are held for others.
Licensing your data is not selling your business
A license covers a copy of your records, not the business itself. Before you sign, ask the firm to confirm in writing what the buyer receives, what it may do with the copy, whether the license is exclusive and how long it lasts.
One point deserves attention: as one law firm put it, once data is used to train a model, "it cannot practically be pulled back out".
What company data deals look like
No one can tell you what your records are worth before a review. MIT Technology Review called pricing "highly uncertain", with buyers' figures varying widely with their estimates of how many useful tasks the data yields.
What has been reported publicly:
- Logistics examples: a 100-person freight brokerage at a reported US$650K and a third-party logistics business at a reported US$500K. Reported by a data licensing firm that pays us referral fees, October 2026. Its figures have not been independently verified. These are other companies' deals, not ours, and they don't predict what your company would be offered.
- Startup archives: Forbes reported that a company sold 13 years of Slack, Jira, email and files for "hundreds of thousands of dollars", and that one wind-down platform's sales were typically $10,000 to $100,000. MIT Technology Review reported a founder receiving about $25,000 for a company's code.
- An airline's internal records: in August 2026, Google won a bankruptcy auction for Spirit Airlines' internal business records with a $10 million bid, according to Bloomberg Law. The records were reported to include about 100 million emails, with personal data excluded from the start. As of October 17, 2026, the court had not approved the sale, and the hearing had been adjourned to November 6, 2026.
What seems to drive value: one firm that values startups' data told MIT Technology Review it looks at headcount, years in operation, data volume and how the team used its tools. That is why we start by asking for your team size and years in business.
What to ask for before you agree
Before you agree to anything, ask to see:
- A review of what you hold, ideally starting with a sample.
- A written offer with the terms of the license.
- A clear description of how personal details are removed before anything is delivered.
- A payment schedule in writing.
The airline sale shows the kind of safeguards a buyer may be asked to accept. As reported, the proposed terms require a separate de-identification firm to certify the data before each transfer, require the de-identification to keep links between records intact, and require the buyer to commit publicly not to re-identify anyone. Even with those safeguards, unions, vendors and members of Congress objected, and the sale was delayed for months. Members of Congress warned that "conventional de-identification safeguards may not be sufficient to protect employee privacy".
What to check before you say yes
Who can decide
You need to own the business or have the authority to decide what happens to its records. Freight agents should check their agreement with the broker to see who controls the transaction records and the software. If the business is in a formal insolvency process, the owner may no longer be the one who decides: in a US Chapter 7 bankruptcy, the company must hand its records to the trustee (11 U.S.C. 521).
What your contracts say
Holding records doesn't mean you can license them. A law firm's October 2026 article notes that email and chat contain other companies' confidential information, often covered by NDAs and commercial agreements. It adds that removing names deals with whether a record can be linked to a person, not whether its content is confidential, and suggests leaving high-risk data out entirely.
Check your software terms too. Some load-board, telematics and WMS providers restrict how their data is used, and some can delete customer data soon after a subscription ends.
Rules for your type of business
Some US rules protect shipment and client information itself, not just people's names:
- Motor carriers and brokers may not disclose a shipment's nature, quantity, destination, consignee or routing without consent where it could harm the shipper or reveal its business to a competitor (49 U.S.C. 14908).
- A parallel rule covers common carriers, marine terminal operators and ocean freight forwarders (46 U.S.C. 41103).
- Customs brokers' client records are confidential; beyond the client, its surety on a particular entry, DHS and other accredited US officers or agents, and a subpoena or court order, they can be disclosed only with the client's written authorization (19 CFR 111.24).
Personal data
Logistics records can hold personal details, and the UK's Information Commissioner's Office says information about employees, sole traders and directors can be personal data where they are identifiable.
- UK: the ICO says that turning personal data into anonymous information is itself processing personal data, and that pseudonymized data is still personal data.
- California: the state's privacy law applies to employee and business-contact data held by the businesses it covers; amendments that took effect on January 1, 2023 ended the earlier exemptions. Data counts as de-identified only if reasonable measures, a public commitment not to re-identify, and contracts binding every recipient are all in place.
- US: staff in the Federal Trade Commission's Office of Technology have written that hashing doesn't make data anonymous.
Ask any firm exactly how personal details are removed, by whom, and whether you can review a sample first.
Questions to ask a licensing firm
- Do you act for me or for the buyer, and how are you paid?
- Who will use the data, and can they pass it on?
- Is the license exclusive, and for how long?
- What exactly is included, and what is left out?
- How is the copy made: do you need access to our email or systems, or do we export it ourselves?
- Who removes personal details, how, and can I review a sample before I sign?
- Does the contract ban re-identification?
- Will the buyer, or anyone else, know my company's name, and will my customers be told?
- What is the payment schedule, and what could change it?
- How much of my team's time will this take, and how long until payment?
- Do I need consent from customers, clients or anyone else first?
- What happens if I say no?
How Insightlanes.com helps
We introduce owners of logistics businesses to a data licensing firm. Fill in our short form with your team size and years in business. Rough numbers are fine. If it looks like a fit, we introduce you to a firm that represents owners in these deals and email you its name. The firm explains its process and terms, usually over a few video calls, and you decide whether to go ahead.
- No cost. If you sign, the firm we introduce you to pays us a referral fee. You never pay Insightlanes.com anything.
- No calls from us. Everything from us is in writing, so you can decide in your own time. If you go ahead, the firm we introduce you to usually explains its process over a few video calls.
- You decide. Nothing is shared without your OK.
Read what you would be agreeing to, and get your own advice before you sign. This guide is general information, not legal advice.
Questions about selling or licensing your company's data
Is it legal to license my company's records?
It can be, but it depends on your contracts, the rules for your type of business and privacy law. A law professor quoted by MIT Technology Review notes that no US federal law bans these sales, but the contracts and rules described in this guide still apply. Get your own legal advice before you sign.
Do I keep my data?
Check the license terms. A license covers a copy of your records, so ask the firm to confirm in writing that your originals stay with you and what the buyer may do with the copy. As one law firm puts it, once records are used to train a model, they can't practically be pulled back out.
How much could my records be worth?
No one can say before a review. Reported figures range from tens of thousands of dollars for some startup archives to a $10 million winning bid for a bankrupt airline's records (not approved by the court as of October 17, 2026), and pricing is still described as highly uncertain.
Does removing names deal with confidentiality?
Not on its own. Removing names deals with whether a record can be linked to a person, not whether its content is confidential.
Do I have to pay anything?
No. If you sign with the firm we introduce you to, it pays us a referral fee out of its own share, so it doesn't reduce what your company is paid. You never pay Insightlanes.com anything.
My business has closed. Can its records still qualify?
They may, as long as you have the authority to decide what happens to them.
Sources
- Forbes, April 16, 2026: labs exhausted public internet data; agentic AI; around 50 simulated-workplace startups; linked records valued over standalone documents; a 13-year archive sold for "hundreds of thousands of dollars"; typical wind-down sales of $10,000 to $100,000
- MIT Technology Review, October 13, 2026: demand "emerged suddenly"; buyers seek judgment calls; no published tests of the benefit; pricing "highly uncertain"; valuation factors; about $25,000 for code; no federal law bans such sales
- Fast Company, April 2026: what AI agents need from documents and emails
- Sierra, June 20, 2024: AI agent benchmark on airline rebooking tasks
- Bloomberg Law (subscription), August 17, 2026: the $10 million winning bid, the records offered and the exclusion of personal data
- Stretto case update, October 2026: the hearing adjourned to November 6, 2026; objections from unions and vendors
- Implicator.ai, August 2026: de-identification preserving links across the data
- The Next Web, October 2026: certification before each transfer; public commitment not to re-identify
- TravelPulse, October 2026: lawmakers' warning on de-identification
- Frankfurt Kurnit via Mondaq, October 6, 2026: third-party confidential information; de-identification versus confidentiality; data can't be pulled back out of a trained model
- 49 U.S.C. 14908: motor carrier and broker disclosure rule
- 46 U.S.C. 41103: ocean shipping disclosure rule
- 19 CFR 111.24: customs broker client records
- 11 U.S.C. 521: records handed to the trustee in bankruptcy
- ICO, Introduction to anonymisation: anonymizing personal data is itself processing
- RPC, September 2025: under the ICO's guidance, pseudonymized data is still personal data
- ICO, What is personal information: information about employees, sole traders and directors can be personal data
- Samsara platform terms and ShipHero terms of service: providers that can delete customer data after a subscription ends
- Fisher Phillips, US state consumer privacy law chart (laws effective January 1, 2026): California's law applies in the employment and business-to-business context; its CPRA amendments took effect January 1, 2023
- California Civil Code 1798.140: the definition of de-identified data
- FTC Office of Technology blog, July 2024: hashing doesn't make data anonymous
- FONASBA Standard Liner and General Agency Agreement (sample): agency files belong to the principal
- Penningtons, July 4, 2024: SHIPMAN 2024 and vessel data
- Acquisition Stars, April 18, 2026: customer data in 3PL contracts