Building a native iPhone app on Genesys Cloud: verified messaging, live video and location

Most organisations reach their customers' phones through someone else's channel: a web chat squeezed into a mobile browser, a messaging app they do not control, or a phone number. Each works, but each starts the conversation knowing very little. The customer has to prove who they are, describe where they are, explain what they can see, and repeat it all if the conversation moves to another day or device.

A native app changes the starting point. The phone already knows who its owner is, where it is, what its camera can see and what state it is in. Built on Genesys Cloud messaging, an app can carry all of that into the conversation, with the customer's consent, while Genesys Cloud keeps doing what it does best: routing, history, recording rules and reporting. We wanted to prove how far that goes, so we built the app for our own staff first, as a proof of concept and a proving ground, and ran it end to end against our Genesys Cloud organisation.

The person taps one button, enters their mobile number and types the six-digit code that arrives by SMS. That is the only sign-in they ever do, and there is no password to forget or leak. The code is sent from the organisation's own number through the Genesys Cloud Agentless SMS API, so no separate SMS provider is involved, and the number it proves becomes the person's identity everywhere that follows.

Behind the button sits a small OpenID Connect provider that we host. Genesys Cloud authenticated messaging is configured to trust it through the OpenID Connect Messenger Configuration integration, so the app can hand Genesys a one-time authorisation code on the person's behalf. Genesys exchanges it exactly as it would for a browser sign-in. The result is one sign-in and two trusted sessions: one with our service and one with Genesys Cloud.

Because the session is authenticated, Genesys Cloud keeps the conversation history with the identity rather than with the device, and with identity resolution switched on it links the conversation to the person's external contact record. Close the app, come back tomorrow or pick up another iPhone, and the same thread is there. The agent sees a named, verified person with their history one click away, not an anonymous guest.

The conversation arrives in the agent's normal Genesys Cloud workspace, routed to the usual queue, with a script popped beside it. A green banner confirms the person is verified, with the method and the time; an amber banner says when they are not. If the person has chosen to share it and their phone reports they appear to be driving, a red banner tells the agent to keep it brief or offer to call back. None of this needs a new agent desktop: it is an inbound message flow, a data action and a script.

The script separates two kinds of fact, and we think every mobile channel should. Trusted facts come from our service and cannot be faked by the phone: which iPhone is registered to this person and when they last proved their number. Reported facts come from the app: device and app version, battery and Low Power Mode, network type, language and local time, accessibility settings such as VoiceOver or larger text, and which permissions are granted. An agent who can see that the person is on 9% battery with VoiceOver on answers differently, and better, than one who cannot.

From the chat, the person can send a photo from their library or take one with the camera. Photos are resized on the phone and sent through the Messenger Transport SDK's attachment support, so Genesys Cloud stores them with the conversation like any other attachment. A pending photo can be removed before it is sent, and it travels with the next message rather than on its own.

Share my location takes a single, accurate position, only when tapped; the app never tracks in the background. It sends the coordinates and their accuracy, a map link and a what3words address, three words that an agent can read back over the phone or pass to a field engineer. The same values are attached to the conversation as structured data, so a flow, a report or an integration can use them.

Share live video turns the phone into a camera for the agent. The phone streams up to 1080p to a video server that we host ourselves, and a private link appears in the conversation for the agent to open in their browser. The person can switch cameras, pause, or take a sharp still photo that drops straight into the chat. When they end the share, the link stops working. Nothing is recorded, and no third-party video service sees the picture.

Anything sensitive is opt-in. A screen in the app called What we share lists exactly what travels with each message, and the two most personal items, telling the team if the person appears to be driving and attaching their position to every message, are switched off until the person turns them on. Each phone permission is requested only at the moment its feature is first used, with plain wording about why.

The app stores no messages. Genesys Cloud is the system of record for the conversation, its history and its attachments, so retention, reporting and access control stay where the contact centre already manages them. Our companion service keeps only what it must: the identity provider's sessions, the registered device, and the latest device context, which the agent's script ignores once it is more than a short while old. Secrets such as integration credentials and the what3words key live on the server, never in the app.

Several findings cost us time and will save it for anyone building something similar. In our testing, a Messenger deployment restricted to a list of allowed domains refused the native SDK, because a native app has no web origin; Genesys's documentation explains that restricted deployments reject API requests from other domains. With auto-start switched on, the conversation and the inbound flow began when the app connected, not when the first message was sent. And custom attributes sent with a message reached the conversation after the inbound flow had already run, too late for a screen pop, so the flow fetches the same context from our service through a data action instead.

On the identity side, Genesys Cloud expects the Authorization Code Flow and the openid scope, and uses the sub claim to identify the person. In our testing, an identity provider that advertised the authorisation-response issuer parameter caused authorisation from the iOS SDK to fail, because the SDK could not pass the issuer back; we stopped advertising it. Scripts need the Message property enabled before a message flow can pop them, and an Agentless SMS send is refused while the number has a connected SMS conversation, though an open web messaging conversation does not block it.

We built the app as our own proof of concept, in steps that were each proved end to end before the next began: identity, sign-in, native chat, rich media, the verified-caller experience, caller context, live video, and distribution to our staff's iPhones. It follows our usual delivery method, from design decisions recorded up front to testing on the simulator, in a browser and on physical iPhones, and it runs on our own infrastructure today.

The same design is a blueprint for what a contact centre can offer its own customers. An insurer could let a policyholder show the damage on live video and drop a what3words pin for the assessor. A utility could take a fault report with the exact location and a photo of the meter. A bank or a public service could start every mobile conversation with a verified customer and their history, without passwords. In each case Genesys Cloud stays the contact centre, and the app becomes the most capable front door it has.

There is more to build, and we are open about it. Replies reach the person only while the app is open today; Genesys documents push notifications for the Transport SDK through Apple's and Google's push services, and that is our next step. Live video currently needs the video server to be reachable directly from the phone's network. A customer-facing version would also be distributed through the App Store and, for most organisations, built for Android too, which the same Genesys SDK supports.