AI Disclosure

Exactly where AI is used in our service, where a human takes over, and what we never do.

Last updated · January 2026

The short version. We use AI heavily in the back office (drafting messages, summarizing chats, briefing techs). We use AI conversationally on the front end (our concierge, photo-to-quote). Every customer-facing message is human-approved. We never train external models on your data.

Where you'll meet AI

SurfaceWhat it doesHuman in loop
Concierge chat Scopes your job, gives a price range, suggests a slot. Runs on a language model we host locally. Always one tap away - press "Talk to a person" any time.
Photo-to-quote You attach a photo; a vision model identifies the install scenario; the concierge uses that to scope. Final price confirmed by your assigned tech on arrival.
Follow-up SMS If you don't book, we send up to two follow-ups. AI drafts; we approve every send. Every message reviewed by a human before it goes out.
Tech pre-brief 30 min before arrival, your tech gets a brief summarized from your chat + photos + notes. The tech reads it before they knock on your door.
Post-install check-in 48 hours after the job, an AI-drafted check-in goes out. If you flag a problem, a human calls you. All replies routed to a human within hours.
Predictive maintenance (robotics) ML models flag wear in autonomous mowers and industrial cells before they fail. Fleet engineer reviews the alert and dispatches.

Models we use

We don't lock our customers into a specific AI vendor. The models we use change as the technology improves. As of the date above:

What we never do

Labeling we use

Real footage - unedited camera capture of a real job.

AI visualization - a generated clip, concept render, or cutaway. Labeled wherever it appears.

Opting out

You can opt out of the AI concierge at any time. Email hello@infrastruqture.com or call (512) 649-4567 and we'll handle everything human-to-human.

Questions about a specific AI interaction?

If a concierge message felt off - wrong price, weird tone, missing context, anything - please let us know. We review flagged conversations, fix what we can, and use the feedback to tune our prompts. AI improvement is downstream of customer signal.

Sample disclosure. This page is intended to satisfy emerging AI-transparency expectations (Colorado AI Act, California AB 2013, the EU AI Act, FTC AI guidance). Exact obligations vary by jurisdiction and service mix. Have an attorney review before going live.