Banalytics Live Chat: Bring Operational Media into the Call
A browser conversation that can include people, live equipment cameras, processed video, and historical playback in one temporary operational workspace.
Many operational calls fail because the people discussing a site do not share the same evidence. One person describes what is happening, another searches for a camera, an engineer opens telemetry in a separate system, and the decision maker tries to reconstruct the situation from disconnected windows. Banalytics Live Chat lets an authorised operator bring the relevant infrastructure media directly into the conversation.
A call becomes an operational workspace
Banalytics Live Chat combines ordinary browser audio, video, screen sharing, text chat, and device selection with something more specific to physical operations: media-capable Things and Tasks from connected Banalytics environments can be dragged into the call. A source can be a live camera, a media task that processes video, or playback of a historical recording.
This changes the role of a call. Participants do not have to rely only on what a laptop camera can see or what a field engineer can describe. With the required access, the operator can place the relevant production area, property camera, processed scene, or incident recording beside the human participants and keep that context visible while the team speaks.
What the design is intended to achieve
- Reduce the time spent moving between a call, dashboards, cameras, and playback tools.
- Let a remote specialist inspect the same operational evidence as the person at the site.
- Keep sensitive side conversations separate from the wider group when necessary.
- Use direct WebRTC channels for media instead of turning the portal into a central video archive or conference-media proxy.
Remote decisions are expensive when context is fragmented
When a meeting concerns a physical asset, the conversation itself is rarely the scarce resource. The scarce resource is trustworthy, current context. A field engineer may have a phone camera but not the right angle. A manager may have dashboard access but not the live scene. A vendor specialist may know the equipment but lack permission to enter the site. Historical footage may exist, yet nobody can place it next to the current condition while the incident is being discussed.
The usual workaround creates delay and ambiguity: participants exchange screenshots, send clips after the call, ask someone to walk back to the equipment, or grant broader system access than the immediate task requires. Every hand-off creates a chance to lose timing, source identity, or the distinction between a live condition and an older observation.
An infrastructure-aware call keeps the evidence attached to the decision. The operator chooses the source, the Agent retains responsibility for acquiring it, and the call gives the authorised participants a common view without moving the entire operational system into a meeting service.
Four situations where the infrastructure belongs in the conversation
Customer or investor site tour
A plant leader wants to demonstrate real operating capacity without stopping production, arranging travel for every participant, or walking through restricted areas with a phone. The call can combine the presenter, selected fixed cameras, a processed quality-control view, and a short historical playback that explains a completed production stage. The result is a guided, evidence-based tour rather than a slide deck that asks the audience to imagine the operation.
Property presentation and remote inspection
A real-estate team often serves clients who cannot visit immediately, while exterior, common-area, construction-progress, or facilities footage already exists in authorised property systems. During the call, the agent can present the property, add permitted live cameras, and open historical interior or renovation recordings when they are relevant. The client can ask questions against a specific scene instead of comparing links received across several messages.
Cross-border production-line support
An equipment vendor may have the deepest product knowledge while the installed line is thousands of kilometres away and physical access is delayed by travel, permits, or production rules. The local support engineer can bring the specialist into a call, show the immediate scene, and add camera or processed media sources from the affected line. The remote specialist sees the evidence needed to guide checks while local personnel retain control of equipment access and any physical intervention.
Operations review and incident reconstruction
Daily planning and incident reviews lose value when status is reduced to verbal summaries. A team can add current cameras for live operating context, processed views that highlight relevant conditions, and recording playback from the incident window. Participants can compare what is happening now with what happened before, agree on the evidence, and assign follow-up work without rebuilding the timeline after the meeting.
Use one common room or a private operational side channel
Group chat
A participant joining through a group invitation becomes part of the common call, regardless of which current participant issued the link. The participant can see and hear the group and receive the media resources that group participants publish into the shared conversation.
Anonymous chat
An anonymous invitation creates a direct side channel to the person who issued it. The anonymous participant sees and hears only that inviter and receives only the inviter's own resources that are being shared into the group. Other participants, their cameras, and their infrastructure media are not disclosed to the anonymous connection.
The anonymous mode is useful during a live operational escalation. For example, an on-site engineer can remain in a management group call while speaking privately with a vendor specialist. The specialist can see the engineer and the media the engineer is already presenting, but does not automatically enter the management discussion or discover the other participants and their sources.
This separation is a topology rule, not only a visual label. The anonymous invitation authorises only the direct connection to its issuer, and group peer lists are not sent across that channel. Anonymous participants are also placed in a separate right-side section in the portal so the operator can distinguish the private side channel from the shared room.
The portal helps peers meet; it does not become the meeting media path
The Banalytics portal performs a temporary switchboard role. It authenticates the active browser session, creates a short-lived invitation, validates which browser sessions that invitation may contact, and exchanges the WebRTC signalling needed to establish the connection. Once peers have a viable path, participant media, screen sharing, text data, and added infrastructure tracks use the WebRTC connection.
The portal is not used as a conference recording store and does not proxy the established media as part of the normal direct path. WebRTC protects media and data in transit. In networks where NAT or firewall policy prevents a direct path, an explicitly configured TURN service may be needed to relay encrypted WebRTC packets; the deployment owner controls that network choice and its credentials.
Adding an infrastructure source does not bypass Agent permissions. The browser still needs an active connection to the relevant Agent, the referenced component must expose a media stream, and the connected account must have the required media permission. The Agent remains the endpoint that acquires live or historical media; the call publishes a selected track for the duration of the collaboration.
From opening a session to sharing operational media
The following sequence uses the current portal and guest-call interfaces. It shows the complete workflow in the same order an operator encounters it.
1. Open Live Chat from the portal header
The phone icon opens the session action without replacing the current operational context. The operator can start a call while the Agent tree and dashboards remain available.
2. Verify devices and choose the room mode
The preview confirms the selected camera before the session starts. Camera, microphone, and speaker are selected independently. The operator chooses Group chat or Anonymous chat and can optionally enter an existing session identifier instead of creating a new room.
3. Work in a call layout designed for operational sources
Human participants remain in the top strip. A separate vertical source panel on the left is reserved for infrastructure media, while the large central stage presents the selected stream. Chat can remain compact or be pinned beside the stage.
4. Choose the appropriate invitation path
The connection menu exposes the session identifier plus two short-lived link types. A group link admits the guest into the common room. An anonymous link connects the guest only to its issuer and keeps the guest in the dedicated private section.
5. Let an external participant join from a browser
The guest opens a mobile-friendly page using the invitation token and grants browser camera and microphone permission. A Banalytics portal account is not required for this temporary guest flow; the token limits where the public WebSocket and WebRTC signalling may connect.
6. Confirm the participant and room topology
A group guest appears with the other participants and receives the common room. An anonymous guest appears only for the inviter in the separate right-side anonymous section and does not receive the rest of the group topology.
7. Share a screen or application window when needed
Screen sharing remains useful for dashboards, documents, and applications that are not represented as media-capable Agent components. The shared screen becomes a selectable participant stream in the guest interface.
8. Drag a camera or media task from the Agent tree
The operator drags a media-capable Thing or Task into the source panel. For a Thing, the chat resolves a suitable media task; for a Task, it verifies that the task exposes media. The existing browser-to-Agent channel acquires the source and the call negotiates a dedicated track for it.
9. Keep the written coordination beside the evidence
Text chat can be expanded or pinned to the side of the main media stage. Messages remain associated with the active call, which is useful for part numbers, short instructions, timestamps, and decisions that should not be lost in spoken conversation.
10. Present the infrastructure source in the guest call
The added camera appears as its own source rather than being flattened into the presenter's webcam. The guest can select it on the central stage, preserving source identity and allowing several operational views to coexist in the same call.
11. Control each participant and source independently
Participant cards expose focused controls for screen sharing, camera, microphone, map, device configuration, and disconnection where applicable. Infrastructure source cards preserve their own identity and can be removed from the broadcast without ending the human call.
Live, processed, and historical media can answer different questions
| Source | Question it helps answer | Example |
|---|---|---|
| Live equipment camera | What is happening at the asset now? | An ONVIF or network camera covering a production cell, loading area, building entrance, or property exterior. |
| Processed media task | What does the operational algorithm highlight? | A video task that adds watermarks, performs classification, focuses a region, or otherwise prepares a reviewable stream. |
| Historical playback | What happened before the current condition? | A stored recording replayed through Recording Playback during incident review, training, or a customer demonstration. |
| Screen or application window | What context exists outside the Agent media model? | A dashboard, maintenance document, vendor application, or diagnostic window shared by a participant. |
These sources should be selected by purpose. A customer tour may need two fixed cameras and one historical clip; a support escalation may need one close operational view and a processed stream; an incident review may begin with playback and then switch to the current live scene. More sources are not automatically better: every track consumes uplink, decoding capacity, and participant attention.
Start with one decision that currently suffers from missing context
- Define the decision. State who needs to decide what, how quickly, and which physical evidence changes that decision.
- Choose the minimum sources. Identify the one or two live, processed, or historical streams that answer the operational question.
- Confirm access ownership. Grant only the component and media permissions needed for the operator who will add the source.
- Select the invitation topology. Use Group when the guest belongs in the common room. Use Anonymous for a direct side channel to the inviter.
- Test the real network path. Verify camera acquisition, browser permissions, direct connectivity, optional TURN policy, uplink capacity, and recovery after a brief interruption.
- Rehearse removal and closure. Confirm that operators can remove a source, disconnect a participant, leave the call, and avoid retaining an invitation longer than intended.
- Keep the system of record separate. If an incident requires durable evidence, retain it through the approved recording, event-history, ticketing, or maintenance workflow rather than treating the call itself as the archive.
Temporary collaboration should not require broad infrastructure exposure
- Short-lived invitations. Guest links are time bounded and resolve to an explicit call topology.
- Scoped signalling. An anonymous token can signal only to its issuer; a group token is limited to the authorised group participants captured for the call.
- No automatic Agent access for the guest. The portal operator acquires and publishes selected infrastructure tracks. The guest does not receive credentials or a general Agent console.
- Direct media design. The portal coordinates establishment of WebRTC channels but is not the normal participant-media path or a call recording store.
- Browser consent remains visible. Guest camera, microphone, and screen access still depend on browser permission and participant action.
- Local responsibility remains local. The deployment owner defines the lawful purpose, participants, camera placement, retention, access policy, and whether a TURN service is permitted.
A secure collaboration design is not achieved by hiding the infrastructure from the people who need it. It is achieved by exposing only the specific evidence required for a defined task, for a bounded time, through a connection topology the operator can understand and close.
Bring the right operational evidence into the next call
Start with one remote support, customer demonstration, property review, or incident workflow and identify the smallest useful media set.