Home services and trades

Can your field service software vendor use your data to train AI?

By SourceX Editorial · Reviewed by Noah Loul · October 9, 2026

Short answer

Whether a field service software vendor can train AI on your data depends on its contract terms, not on who owns the data. Terms typically let the vendor use customer data to run the service; the real questions are whether they add product improvement, aggregated data or model training rights, and whether you can opt out.

Key takeaways

Owning your data does not settle the question; the license you granted the vendor decides what it may do.
Product improvement and aggregated data clauses are where AI training rights usually sit.
AI feature terms, order forms and change notices can override or extend the main agreement.
Your vendor's rights and your own right to export and license your records are separate questions.
Renewal is the best moment to negotiate opt-outs and full export rights into the order form.

Can your field service software vendor train AI on your data?#

Your field service software vendor can train AI on your data only to the extent its contract allows, and that wording differs between vendors and between versions of the same vendor's terms. Ownership clauses rarely settle it, because a contractor can own its data while granting the vendor a broad license to use it.

Three permissions matter. The first is use to provide the service, which every platform needs. The second is use to improve or develop the vendor's products, which may or may not extend to model training. The third is the right to create aggregated or de-identified data, which some terms treat as the vendor's own once created.

This is general information, not legal advice. Contracts differ, so review your own agreement with counsel before relying on any reading of it.

Where the answer sits in your contract stack#

Your vendor's AI training rights sit across several contract documents, and the most specific or most recent one often controls. Field service platforms commonly publish a master subscription agreement or terms of service, then layer on order forms, a data processing addendum, a privacy policy and, increasingly, separate terms for AI features.

Download a dated copy of each. Terms pages often change without a public version history, and the wording in force when a record was created may matter later.

Master subscription agreement or terms of service: ownership, license grants and the vendor's right to change terms.
Order form: negotiated changes that, under many agreements' order-of-precedence clause, override the standard terms for your account.
Data processing addendum: how personal information is handled on your behalf and which subprocessors receive it.
AI or product-specific terms: rules for features such as call summaries, smart dispatch or estimate suggestions.
Privacy policy and trust pages: public descriptions of practice, usually not the binding contract.
Change notices: emails or in-app notices of updated terms that may take effect on renewal or continued use.

Clause checklist for AI training and data use#

A clause checklist turns a long contract into short answers you can compare across vendors and renewal cycles. Read each clause with one question in mind: what may the vendor do with job records, technician notes, photos and call recordings, and can you stop it?

Clause checklist for AI training and data use
Clause	What to look for	Question to put to the vendor in writing
Ownership of customer data	Whether customer data, including technician uploads, is defined as yours	Does customer data include photos, recordings and notes created in your mobile app?
License to the vendor	Scope limited to providing the service, or extended to improving and developing products	Does improving the services include training machine learning models?
Aggregated or de-identified data	Whether the vendor may create derived data and who owns it	What de-identification method do you use, and can derived data reach third parties?
Usage, service or telemetry data	Whether the vendor defines data about how you use the platform as its own, and how broadly	Does usage data cover only clicks and logins, or also job content, notes and attachments?
AI training and model improvement	Explicit statements on training vendor or third-party models	Is our data used to train models that serve other customers?
Opt-out	An account setting or contract term that switches training off	Where is the opt-out, and does it cover data already collected?
Third-party AI providers	Subprocessors that receive data for AI features	Which providers process our data for AI features, and may they retain it?
Export on termination	Format, scope, fees and the window to retrieve data	Can we export all jobs, notes, attachments and recordings with their IDs?
Deletion after termination	When data is deleted and whether derived data survives	What happens to derived or aggregated data after we leave?

What aggregated and de-identified mean in field service records#

Aggregated and de-identified data are records stripped of direct identifiers or combined across many customers, and vendor terms usually leave the method to the vendor. In field service, that can mean statistics about job durations or parts usage, but it can also mean individual technician notes or photos with names removed.

The difference matters because free-text notes, photos and call recordings are hard to de-identify reliably. A note that mentions a gate code, a dog's name and a cross street can point to one household after the customer name is gone. Ask how the vendor treats free text and media, not only structured fields.

Also ask whether derived data can describe your company as a business. Aggregates across a large customer base reveal little about one contractor; job-level records reveal pricing, close rates and technician practices you would treat as confidential.

Your vendor's rights and your right to license are separate questions#

Your vendor's rights and your own right to license your records are separate questions, and owners often blur them. A vendor's permission to use data for its products does not usually stop you from licensing your own records, but your contract can limit how you get them out.

Check the API and export terms. Some platforms restrict bulk exports, automated extraction or use of API data beyond running your business, and some charge for full-history exports. Confirm that you can export job history with attachments and IDs, and that nothing restricts what you do with your own exported records.

Disclose existing vendor rights if a buyer ever asks for exclusivity. An exclusive license to your records generally cannot take back rights you already granted your software provider, so those rights belong in the rights review from the start.

Illustrative: an electrical contractor reviews its terms at renewal#

Illustrative: a fictional electrical service contractor with a residential service department and a small commercial team runs dispatch, estimates and invoices in a field service platform and stores panel photos and voice notes in the mobile app. Renewal is approaching, and the vendor has announced new AI features.

The operations director pulls the current terms, the AI feature terms and the signed order form, then sends five written questions drawn from the checklist. The replies confirm the company owns customer data, that product improvement can include model training on de-identified data, and that an account-level opt-out exists for some features.

The owner switches on the opt-out, negotiates a full export with attachments into the renewal order form, and files dated copies of every document. The company now knows what rights the vendor kept and that it can export its own history for a separate licensing project.

What to do before your next renewal#

Renewal is when a contractor has the most leverage over its field service vendor, so prepare the data-use questions before the quote arrives.

Download and date every current contract document, including AI feature terms.
Mark the license grant, aggregated data and AI training clauses.
Send the checklist questions to your account manager and keep the written replies.
Find and record opt-out settings with screenshots and dates.
Run a test export of closed jobs with notes, photos and recordings.
Put negotiated changes in the order form, and confirm in the order-of-precedence clause that it controls over the standard terms.

How SourceX reviews platform terms#

SourceX reviews platform terms during the Rights step of the SourceX five-step transaction, alongside customer agreements, privacy notices and any franchise documents. The aim is a clear statement of what the contractor may license and what its vendors already hold.

Findings are recorded in the SourceX Evidence Packet under licensing rights and permitted use, so a buyer can see that exports were taken under terms that allowed them. SourceX's dataset rights, including any training use, are set out in the signed supplier agreement, and it needs no access to a vendor account during the fit check.

Frequently asked questions

Can I stop my vendor from training AI on data it already has?

Sometimes. Opt-outs often apply going forward, and the terms may say nothing about models already trained or derived data already created. Ask the vendor how an opt-out treats existing data and get the answer in writing before relying on it.

Does my vendor own the photos my technicians upload?

Usually not, if the terms define uploads as customer data, but check that definition. Even when you own the photos, the license you granted may let the vendor use them for product purposes. Ownership and use rights have to be read together.

What if the vendor changes its terms mid-contract?

Many terms let the vendor update them with notice, with changes applying at renewal or on continued use. Watch for notices, save each version, and raise objections before renewal, when you can negotiate an order form that fixes key data-use terms.

Do AI features like call summaries send data to other companies?

They can. AI features often rely on third-party model providers listed as subprocessors. The data processing addendum and the subprocessor list show who receives data, under what conditions, and whether those providers may retain it or train on it.

Should I switch vendors over AI training terms?

Rarely on terms alone. Switching carries real migration risk to job history, notes and attachments. Most contractors get further by negotiating an opt-out and export rights at renewal, and by keeping a full independent export of their history.

Related resources

See if your company qualifies

A short company assessment. No data uploads are needed.

Use the software vendor terms checklist Talk to SourceX