Why we built Nexus AI on top of Smallworld GNM instead of around it
Why we built the Nexus AI Platform on top of Smallworld GNM instead of around it.
NEXUS AI PLATFORM
Nexus connects the systems a utility already runs, centralizes what they hold, and adds AI that answers across them and acts in the system of record — on the models you choose, inside your own infrastructure.
Data stays inside your infrastructure. Communication is encrypted end to end, with connected systems and with the model alike. The platform is closed by default: no external access is required or permitted unless you ask for it.
Nexus is not tied to one AI provider. Run the models you choose, on premises or in your own cloud, and reuse subscriptions you already pay for — including different models for different groups of users at the same time.
Functionality ships as self-contained pods, and pods do not depend on one another being installed first. Start with the one or two that answer this quarter's problem and add the rest when the work calls for them. Nothing here needs a platform-wide programme before it does anything useful.
Four steps, in this order. Everything a pod does sits on top of them.
Where a pod has to reach a system outside the platform, an administrator registers that system and connects to it by API or by a direct database connection. The connection and its mappings are defined once, not again per project.
What those systems hold becomes available in one place, to every pod, under the access rules you already apply to your people.
Questions are answered across systems, in plain language, without the person asking needing to know which system holds which field.
The platform writes back to the system of record, inside the rules you set, with what was done and by whom recorded as the write happens.
Describe the dashboard you need in plain language. Nexus builds it from the network records themselves, not from an export someone took last quarter.
Every pod runs on the same platform, under the same access rules, and each one stands on its own. Start with whichever matches the problem in front of you.
Set up your own connections and APIs. Added when a pod has to reach a system outside the platform.
Connect a system once and every pod can use it, so nothing has to be re-integrated project by project.
The ones most clients end up running. Ask, act, orchestrate and report across the connected estate.
Designers ask in plain language and the work happens inside Smallworld, with network integrity checked before anything is posted.
The GIS job queue and the job servers look after themselves, and a failure tells you at the time instead of surfacing days later.
Ask across every connected system in plain language, then share the reports and dashboards that come back rather than exporting them.
Every layer and every connected system on one map in the browser, with plain-language questions over all of it.
One job each, with a simplified map of its own. Added when that job is the one in front of you.
Someone else’s drawing arrives as a file and leaves as a checked project in your GIS.
Drop a pin where the customer is and get back a designed, rule-checked project sitting in your GIS.
Your own team corrects displaced network geometry, and the patterns they build up get reused from one area to the next.
Prune where the network actually fails rather than where the calendar says, with everything from field collection to the finished work order in one place.
The utility's own telecom network — to splice and port level — on the same map as the grid it keeps running, with joint use alongside.
Applied to your environment rather than licensed as separate pods.
An autonomous service desk — for example, integrated with ServiceNow.
A Magik/VMDS engineering workspace.
A crew member asks a question from the messaging client they already use. Nexus answers against the network, not against a document about the network.
Platform overview
What Nexus does and where it sits against the systems you already run. No registration.
Real situations, written out end to end — including what the platform says back.
GIS designers and technicians
A design package lands and someone types it in. Five hundred poles, one at a time, against standards that live partly in a manual and partly in people's heads.
Design and engineering teams
A connection request arrives and a designer picks it up. Find the nearest point of connection, lay in the extension, apply the standards, check it against the rules, build the project, route it for approval.
Engineering teams who own contractors
Contractors, developers and engineering firms deliver designs in CAD, in geospatial exports, sometimes in a spreadsheet of coordinates. Each one has to be brought in, checked and turned into something the GIS will accept.
Planning and asset-management teams
Every budget season somebody rebuilds the same spreadsheet. Asset ages come out of the GIS, unit rates come out of the ERP, and the join happens in a workbook that nobody else can reproduce.
Operations and maintenance managers
Everybody knows the feeder that keeps failing. Nobody can show it.
OFFICIAL PARTNERSHIPS
TRUSTED BY
THE OTHER WAY WE WORK
MagikDev implements and supports the systems Nexus connects to. Smallworld, Electric Office and PNI are GE Vernova products and we are an official partner — the same engineers who deliver those implementations build Nexus.
INSIGHTS
Why we built the Nexus AI Platform on top of Smallworld GNM instead of around it.
Three takeaways from Eurelectric and EY's 2024 Grids for Speed report.
Our team's takeaways from GE Vernova's 2025 Annual Report: $38B revenue, $150B backlog, and what the Electrification supercycle means for Smallworld GNM operators.
A technical session covers your estate, where the platform would run, and which pods come first.
Book a technical session