04 Create
Custom IIoT apps, built by chatting
Say "build an OEE dashboard for our line" and an AI agent first proposes a screen plan, then — after your approval — builds a real application, shows it in live preview, and deploys it in one click. An agentic IIoT app studio that lets field engineers, without writing code, push apps all the way to tablets, kiosks, and control-room TVs.
Chat → App
From conversation to deployed app
A Capabilities · In depth
Type one line in the hero and the agent doesn't code right away — it proposes a screen plan (screens, widgets, required data) as a card. On approval, it generates a real app, verified in live preview across three device frames (desktop, tablet, phone), then deploys in one click. Deployed apps open instantly on field phones via QR.
The agent console on the left and the live preview on the right share one screen. Point at any part of the screen with element selection to request changes, check responsiveness instantly with desktop, tablet, and mobile previews, and finish the whole journey — version history to one-click deploy — inside the builder.
Writing code is only the start. The agent runs type checks and builds on its own, visually inspects the rendered screen, finds the code behind broken UI and errors, and fixes it itself — looping without human intervention until errors reach zero. A field engineer alone gets the experience of having a developer sitting next to them.
A semantic layer that reads the platform ontology resolves "line 3 temperature" to the exact asset and tag IDs — deterministically. The agent doesn't guess IDs. An auto-generated 187-endpoint SDK fetches live data, and data bindings are auto-scanned to show which assets and tags each screen uses, always visible in the cockpit.
Save a veteran engineer's judgment as a skill and the agent pulls it in when relevant (progressive disclosure keeps context light). Turn "tell me if line 3 exceeds 80°C" into a natural-language watcher that monitors on a 1-minute tick and alerts via real-time toast and messenger webhook. Skill packs port between air-gapped plants as files.
Run the enterprise with Studio's own accounts and three roles (admin, builder, viewer), audit logs, usage stats, and white-label branding. Deployed apps extend to private sharing, embed code, an auth-free public gallery (field kiosks), and a TV wallboard (full-screen rotation).
B Architecture
An AI agent plans, gets approval, and builds your chat request, verified in live preview inside an isolated per-session container before deploying. A semantic layer resolves natural language to real asset and tag IDs, and a 187-endpoint SDK connects to live data.
C Use Cases
Outcome
Field-led app development without IT bottlenecks
Outcome
Dramatically shorter app development lead time
Outcome
Unified deployment from shop floor to control room
E More Capabilities
An AI agent edits a real app through a plan-approve-build flow and 3 modes. Fast, while you stay in control.
Fixes itself with type-check and vision, verified in real time across 3 device frames.
An ontology semantic layer resolves NL to real IDs and connects to live data via a 187-endpoint SDK.
Save engineer know-how as reusable, progressively disclosed skills, and port between plants with skill packs.
Run NL monitoring rules on a 1-minute tick with real-time toasts and messenger webhooks.
Versioning and rollback, QR/embed/gallery/TV wallboard, own accounts with 3 roles, audit, and white-label.
F Technical Specifications
G Product Screens
D FAQ
Yes — build apps with natural-language chat. The output is real code, so if you want, you can view it in the code viewer or export it to a remote Git.
They connect to the platform V5 API (assets, tags, alarms, ontology). A semantic layer deterministically resolves natural language like "line 3 temperature" to real asset and tag IDs, so the agent doesn't guess.
Type-check and vision self-fix loops fix errors within the turn. Every turn is a git commit for undo from history, and deployed versions roll back in one click.
All features work with a file fallback (no DB required), and you can point it at an on-prem model (OpenAI-compatible). Skill packs and backups port offline.
Private sharing, embed code (iframe), QR (field phones), an auth-free public gallery (kiosks), and a TV wallboard. Origin isolation and server-side keys keep the security boundary.
A 15-minute demo walks you from ingest to AI. Check the fit for your plant with an expert, right away.
