Remote visual inspection (RVI): how it works in service and claims

This post in 30 seconds.

• Remote visual inspection means a qualified person assesses a physical situation through a camera someone else is holding, without being there. The term also has an older industrial meaning (borescopes inside turbines); this post covers both and owns the second.
• It doesn't need a live call. The person on site records what your flow asks for, when it suits them, and your reviewer looks at it when ready. What makes it an inspection is the required views, the qualified interpretation and the structured record, not a live feed.
• Insurers use it to settle claims without sending an adjuster, over a video call or through self-guided capture sent at first notification of loss.
• Written for service and dispatch leads who decide whether a van goes, and for claims leads who decide whether an adjuster drives to the loss.

Remote visual inspection (RVI) is the assessment of a physical situation by a qualified person who is not there, from photos and video captured by someone who is. A service desk uses it to decide whether an engineer needs to go. An insurer uses it to settle a claim without sending an adjuster. It can run over a video call, or as self-guided capture that the reviewer looks at later.

"It's making a noise and there's a light flashing." That is the whole brief the desk gets, and somebody now has to decide whether an engineer drives forty minutes to look at it. Aquant's 2025 Field Service Benchmark Report puts the share of truck rolls that were never necessary at 14%, one in seven.

On a service desk, remote visual inspection sits between that phone call and that dispatch. Below: both meanings of the term, what separates an inspection from a WhatsApp photo, why it needs no live session, how the asynchronous version works, how service and claims teams use it, and what the record should contain.

We build Venta Capture, VentaVid's guided capture platform, which is one way of running the asynchronous model described here, so read the worked example with that bias in view. Everything else is sourced. If you would rather see the mechanics than read about them, start a free account and send yourself a capture link while you read.

In this post:

• What remote visual inspection means
• What makes it an inspection, not a video
• Does it need a live session?
• How the asynchronous version works
• How insurers use it for claims
• What a structured inspection record includes

Where this guide comes from

The industrial definition comes from the non-destructive testing world (Wikipedia and the equipment makers' own pages). The operational one comes from field-service and claims practice: our own capture flows, public programmes that inspect by video call, UK conduct rules and the benchmark reports the industry quotes. Every number carries its source link; the one Capture figure is our own and labelled as such.

What remote visual inspection means

The "remote" part is the reviewer. The "visual" part is that they judge from images, not from a description. The "inspection" part is that the result is an assessment against known criteria, with a record.

Remote visual inspection is defined by who assesses and what gets recorded, not by the camera being live or by any particular device. That sentence carries most of the post, so it's worth reading twice.

The term has two lives, and search engines mix them up:

• The industrial (NDT) meaning. Borescopes, videoscopes, fiberscopes, push cameras and robotic crawlers that let an inspector look inside a turbine, a pipe or an aircraft engine without dismantling it. Wikipedia classifies it as a branch of non-destructive testing; Evident defines it as inspecting areas "too dangerous, remote, or inaccessible for direct visual inspection". Here the inspector is on site and "remote" means a few metres of probe. A borescope inspection is the classic example.
• The operational meaning. A service desk, an adjuster, a maintenance supervisor or a technical-support engineer sees a boiler, a machine, a rooftop unit or a leak through the phone of whoever is standing next to it. Here the inspector is the one who is remote. Some vendors rename this "remote virtual inspection" to keep it apart from the probe version; most people just say remote inspection or remote video inspection.

This post is about the second meaning. If you need the first, you need equipment, not software, and the equipment makers' pages are the right place.

Within the operational meaning, three routes get the eyes to the problem, and each one solves a different constraint:

1. A physical visit. Solves "someone needs to see it." The default, and the one you're paying for every time a van rolls.
2. A live remote session. Solves "our expert doesn't need to travel." The customer and the expert meet on a video call at the same moment.
3. Asynchronous guided capture. Additionally solves "our expert doesn't need to be there at the same time either." The request goes out now, the person on site captures when convenient, the reviewer assesses later.

Most guides online describe option two and call it the category. This post argues that option three is the one that changes how a service desk runs.

What makes it an inspection, and not "send us a video"

Every service desk already does a crude version of this. "Can you send me a picture?" The customer sends one photo of the front of the unit, taken from the doorway, and the desk learns almost nothing it didn't know from the call.

That is footage. It isn't an inspection. Three things turn one into the other.

• Required views, defined in advance by you. A visual inspection has inspection criteria before it starts. For a boiler that might be the rating plate, the display with the fault code, the flue connection, the pressure gauge, the space around the unit and a short video with the sound of it trying to fire. The person on site doesn't decide what to show; your flow does.
• Interpretation by a qualified reviewer. The camera holder is the eyes. The assessment is made by the person who knows what a scorched terminal or a wrong-sized flue looks like. That split is the whole point: you use the person who's already there, and you keep the judgement with the person who is qualified to make it.
• A structured record tied to a case. An inspection produces an inspection report, or at least a case with the captures, the answers, the time received and the reviewer's conclusion. A photo in a WhatsApp thread produces a memory.

A customer video becomes an inspection the moment your organisation decides what has to be in it and a qualified person judges it against a record. Nothing about that requires the camera to be live.

The reason the required-views step matters so much shows up in the failure data. Comparesoft's summary of first-time-fix research lists five causes of a failed first visit: the wrong parts, no access, the wrong skill level, not enough time scheduled, and a wrong diagnosis. Four of the five are information gaps that existed before the van left.

The minimum evidence set you define for each job type is the tool that closes them, and we've written separately about why open photo uploads keep failing and what a written shot list changes.

Does remote visual inspection need a live session?

No. A live video call is one way to run it, and it is the way most software in this space is built, which is why the assumption is so widespread. But live is a design choice with a cost, and the cost is worth naming.

A public example makes it concrete: the City of Calgary runs remote video inspections for plumbing, gas and electrical permit work. They happen by live video call, Monday to Friday between 8 a.m. and 3:30 p.m., with inspectors calling back from a queue on a first-come basis.

The person on site must be ready to "navigate around the property or structure as required, regardless of weather conditions" with a charged phone. It removes the drive, and it couples two calendars at the same hour on a weekday.

The live model solves travel; the asynchronous model also solves timing, and timing is the constraint that decides whether the thing scales. At ten inspections a week, a callback queue is fine. At five hundred customer-reported faults a week, you would be staffing a call centre of engineers to watch phones.

The asynchronous version breaks the coupling:

• No appointment. The request goes out now; the capture happens when the person is next to the equipment, which might be tonight.
• No app and no account. A link opens the guided flow in the phone's browser. Nobody installs anything to report a fault once.
• No live agent. The guidance is built into the flow (what to show, in what order, which question to answer), so the expert isn't needed while it's being recorded.
• No calendar alignment. The reviewer works the queue when they work the queue. Capture capacity stops being tied to engineer availability.

Live still wins in specific cases. If the reviewer has to steer someone through a manipulation in real time (open this panel, now hold the meter here), if the situation is safety-critical, or if the answer has to be given on the spot, schedule the call.

The honest position is remote-first, not remote-only. The sensible pattern: capture first, escalate to a call or a visit when the capture says you should. The wider category, including the products built around live sessions and the ones built around inspector apps, is mapped in our remote inspection software guide; this post stays on the definition.

If a fault queue with a callback window is exactly your bottleneck right now, book a demo and bring one real job type; we'll build the flow for it on the call.

How asynchronous remote visual inspection works, step by step

The loop underneath every version of this is the same six beats: request, guide, capture, verify, review, act. Here is how it runs when the capture is asynchronous, using Venta Capture as the worked example (bias declared above).

1. Request. The desk sends a secure, personal capture link by SMS, email or WhatsApp, or shows a QR code on a screen, an invoice or a service sticker. The link is tied to the job, so nothing has to be matched up by hand later.
2. Guide. The customer, tenant, operator or contractor opens it in the phone's browser. A short page says who is asking and why, then the steps run one at a time behind a progress bar: a scan of the serial sticker, three photo slots with an example photo and a one-line instruction each, a yes/no question that steers the next step, a video step with an on-screen prompt such as "walk around the unit and describe out loud what you hear". This is the guided capture part, and it is where your engineers' knowledge lives.
3. Capture. Photos and video are recorded in the session, on that device, with the time received, the device context and the answers logged alongside them. If the flow allows an extra upload from the gallery (a photo taken last night, say), it arrives labelled as unverified so the reviewer knows which is which.
4. Verify. The submission is sealed on receipt. The reviewer can see what happened during the session and whether the file is still the file, which matters later if anyone disputes what was shown.
5. Review. The case lands in a shared inbox with a reference number in your own format. Someone claims it, watches the ninety seconds of video, reads the answers and the spoken-explanation transcript, and either decides or asks for a retake of one specific step, in one click. The retake hangs off the original case.
6. Act. Resolve remotely, quote, ship a part, dispatch the right engineer with the right skill, or arrange a physical inspection. Whatever the decision is, it was made with the equipment in view.

When I build a flow for a new job type, the first draft always asks for too little. The fix is to watch what the reviewer has to ask for afterwards and add those steps. Two rounds of that and the retake requests thin out.

On the support side, Venta Capture reports that 23% of customer problems are solved without dispatching a technician at all, because the desk could see enough to fix it or to send the right part. The rest still get a visit, but a visit that was planned against the fault code and the photo of the connection, not against "it's making a noise". That is what remote triage means in practice, and the full guided video capture mechanics are on the product page if you want the detail.

Where service and maintenance teams use it

The same loop carries a lot of jobs. These are the ones we see most in field service and maintenance, each with a link to the deeper post.

• Pre-dispatch triage on reported faults. The customer or site contact shows the fault before anyone is sent. Covered in depth in remote diagnostics without sensors.
• Pre-visit assessment and site readiness. When the visit is going to happen anyway, see the access, the space and the installed kit first, so the right engineer arrives with the right parts. See pre-visit assessment in field service.
• Cutting avoidable truck rolls. TSIA puts the cost at more than $1,100 to roll every truck every day, up from $750 in 2011. What a truck roll costs you and how to run the numbers on your own queue is in reduce truck rolls.
• Pre-quote capture. Repairs, installations and upgrades that normally need a survey visit before a price. The enquiry comes with the captures; the quote goes out without the drive, or the survey is booked knowing what it's for.
• Completion evidence and contractor proof of work. The same flow run at the end of a job instead of the start: show the finished work, the readings, the tidy-up. See proof of work in field service.
• Dealership service intake. The automotive version: the customer shows the noise, the warning light and the tyre before the car is booked in, so the work order starts with something to go on.

Every item on that list is the same six-beat loop with a different shot list, which is why a tool scoped and priced per "inspection" tends to fight you the first time you want to use it for a handover. The benchmark to hold it against is first-time fix rate: Aberdeen Group puts the industry average near 75%, with best-in-class operations at 88% to 90%; the gap is mostly what the engineer knew before leaving.

The field-service use case page walks through the pre-dispatch, site-readiness and completion flows with the actual steps if you want to see one laid out.

How do insurers use remote visual inspection for claims?

In a claims team, RVI means the adjuster assesses the damage without travelling to it, from images captured by whoever is at the loss: the policyholder, a tenant, a driver, a repairer. The adjuster, often a desk adjuster, keeps the judgement on cover, cause and extent; the person on site supplies the eyes.

In some lines it is already the normal route. In US motor claims, photo estimating accounted for 26.4% of inspections on repairable claims in 2025, according to CCC Intelligent Solutions' Crash Course 2026 report, as reported by Claims Journal.

Insurers run it in two ways:

• Live RVI. The adjuster and the policyholder join a video call. The adjuster steers ("closer to the skirting board, now the pipe under the sink") and keeps what the tool records from the call.
• Self-guided capture. At first notification of loss the policyholder gets a secure link, walks through the shots and questions the insurer defined, and submits when they are standing next to the damage. The adjuster reviews the case later.

Which claims suit remote visual inspection, and which still need a visit?

Value thresholds are the usual first filter. Sorting by what the camera can show works better:

The middle column is where RVI pays most: the routing decision waits until the evidence is in. One rule the table cannot hold. A policyholder who cannot record (age, distress, disability, a broken phone) gets a call or a visit without having to argue for it. In the UK that sits under the FCA rule that an insurer must "provide reasonable guidance to help a policyholder make a claim" (ICOBS 8.1.1). A link that tells the claimant what to show is guidance. A link nobody in the household can use is not.

The full routing rule, with a minimum evidence set per claim type, is in desk adjusting: how to settle claims without a site visit.

What does the insurer need in the file?

A field visit answers one question without anyone noticing: the adjuster stood there and took the photos. A remote file has to answer that question itself. Four questions decide whether the desk can settle:

1. What is it? The identifier that ties the evidence to the policy: plate and VIN, a serial plate, a street view of the property, the room.
2. What happened to it? The cause, in the policyholder's own words and in the picture.
3. How bad is it? The extent, from far enough back to see the whole and close enough to see the material, with something for scale.
4. How did this evidence get here? Recorded now or picked from the gallery, received when, from which device, unchanged since.

The fourth is where remote files tend to be thin, and it is the one an ombudsman or a court asks about later. The file should hold:

• The required shots per claim type, plus the policyholder's answers and a transcript of what they said on video.
• Which images were recorded in the session and which came from the gallery, labelled separately. A gallery photo can be legitimate; it proves less.
• The receipt time, verified server-side: it proves when the insurer received the file, not when the loss happened.
• A seal on receipt and an independent timestamp, so anyone can check later that the file has not changed. That is the digital chain of custody.
• The audit trail: link sent, link opened, submission received, who reviewed it, every retake request and the reason given.

In Venta Capture that set arrives as one case: each file sealed on receipt with its own fingerprint, a visible stamp with date, time and seal on photos and video recorded in the session, a timestamp from an independent timestamping authority, and a printable proof of seal. Depending on the workflow, the case can also carry session checks: a spoken code issued for that session, a changing session code in the frame of every photo and video, a motion step that compares the phone's movement sensor with the movement in the image, and a device consistency check. More than 25 control points per submission in all, each an observation for the handler, never a verdict. The remote claim inspection page walks through the claims flow; sealed evidence explains the seal.

Live RVI or self-guided capture: which one fits a claims team?

After a storm, five hundred policyholders can record their damage the same evening; nobody staffs five hundred video appointments. Live RVI still earns its place when the adjuster has to steer or the claimant needs help. The pattern that works is the same as in field service: capture first, escalate to a call or a visit when the file says so. The tools on the market, sorted by which of these models they run, are compared in best remote inspection software for insurance claims.

What a structured inspection record includes

The live-session guides skip this part, because a call doesn't naturally produce one. If the output of your remote visual inspection is a screenshot someone remembered to take, you have footage again.

A structured record, the thing we call a visual evidence package, holds the following, in one case, under one reference:

• The required captures. Each photo slot and video step, labelled with what it was supposed to show, so a reviewer can see at a glance that the rating plate is missing rather than hunting through a gallery.
• The answers. Every conditional question ("Is the unit powered on?", "When did it start?") stored as data next to the media, which is what lets a system route or filter cases instead of a human reading each one.
• A transcript. Spoken explanations from the video step transcribed and searchable, so "it clicks three times then stops" is text in the file, not a memory of a call.
• Time and device context. The receipt time, verified server-side, plus the device and session details. Be precise about what this proves: the timestamp proves when your system received the submission, not when the fault occurred.
• Integrity information. A digital seal with a fingerprint that can be checked externally, and a session timeline, so a sealed submission can stand up when a contractor, a customer or an insurer later disagrees about what was shown.
• Your own case reference. A number in your format, plus a separate field for the reference your ticketing or CRM system already uses, so the record joins the job rather than sitting beside it.
• The activity log. Link created, link opened (recorded server-side), submission received, who claimed it, who it was assigned to, every note, every retake and why. That is the audit trail, and it is what distinguishes a record from a video with a date on it.

A record built this way can be routed, filtered and handed to a ticketing or claims system through an API. A view link lets a colleague without a login see the case, and workflow rules forward a case by status; the workflow automation page shows how those rules are set.

What a record should contain across sectors, and what auditors and adjusters ask for, is the subject of our cross-sector piece on photo evidence standards for inspections.

What remote visual inspection can't do

Four limits, stated plainly.

• It doesn't diagnose. It gives the qualified person better visual access. The diagnosis, the safety call and any certification decision stay with the human who is qualified to make them.
• A genuine recording can still mislead. A video recorded in the session, on that device, is harder to manipulate than a gallery upload, but harder does not mean impossible. Integrity signals are a reason for a human to look closer, never a verdict on their own.
• Not every visit disappears. Aquant's report shows the honest spread: avoidable dispatches run at 3% for top performers and 24% for the bottom fifth. The promise is see first, act second, and make the visits that remain better informed.
• It needs a person on site with a phone and a signal. If nobody is there, or the thing to be inspected is inside a pipe, you're back to the industrial meaning of the term: a drone, a crawler or a scope.

If the term you keep hearing is the other one, remote visual assistance vs remote visual inspection sets the two apart. For the capture side in practice, see guided video capture for field service.

Frequently asked questions

What is remote visual inspection (RVI)?

It is the assessment of a physical situation by a qualified person who is not on site, using images captured by someone who is. In industrial non-destructive testing the term means borescope, videoscope and crawler inspection of components nobody can reach. In field service and insurance claims it means seeing a fault, an installation or a loss through the phone of the person standing next to it.

What do remote visual inspection services include?

That depends on who sells them. NDT service companies send technicians with borescopes, push cameras, drones and crawlers to look inside pipes, tanks, turbines and vessels. A remote visual inspection system in the operational sense is software: a link that guides the person on site through the required shots, and a case where a reviewer assesses them.

Is remote visual inspection the same as remote video inspection?

Mostly, yes. Remote video inspection is the name many permit offices and insurers use for the version run over a video call, which needs the person on site and the expert free at the same moment. Self-guided capture removes that requirement: the person on site records what the flow asks for when it suits them, and the reviewer assesses it later.

How do insurers use remote visual inspection?

To assess claims without a site visit. The policyholder shows the damage, either on a video call with the adjuster or through a guided capture link sent at first notification of loss, and the adjuster decides on cover, cause and extent from the file. It suits visible, single-item and motor damage best; structural, hidden and disputed losses still get a visit.

Can you trust video a customer or policyholder recorded themselves?

More than a gallery upload, and more still when the platform records what happened around it: receipt time, device context, a seal, a session timeline. It is still not proof of the event, only of the submission, and a real recording can show a misleading situation. Treat integrity signals as a reason to look closer and keep the judgement with a person.

What is the difference between remote visual inspection and remote visual assistance?

Remote visual assistance is a live support session: an expert sees through the customer's camera and talks them through a fix in real time. Remote visual inspection is an assessment with required views and a record, and it does not have to be live. Many teams use both: capture first, escalate to a live session when the capture shows they need one.

Talk to us

If your desk still decides whether a van goes on the strength of a phone call, the cheapest experiment is one capture flow on one job type. Pick the fault you dispatch on most, write down the five things the engineer wishes they'd seen first, and make that the flow.

What it takes to start.

• A free plan, no credit card. Build the flow, run a test capture on your own phone, send the first real link the same day.
• Live in 10 minutes for a standard flow. Stuck? Book a free setup call and we build your first flow together.
• Nothing for the customer to install. They get a secure, personal link and a browser page in their language.

Venta Capture is built by the VentaVid team, the people behind personalized video for sales and service teams.

Book a demo and bring a real job type, or start for free and send yourself the first link now.