MODERAN turns multi-vendor alarm-storms into one root-caused incident, blocks bad vendor updates before production, and makes idle edge compute a sovereign AI revenue line — without subscriber data ever leaving your country.
Competitors solve one layer. MODERAN unifies all four under a shared model — installed inside your own network, on your terms. Click a layer to see how it works.
Every vendor, one console. Alarm-storms become a single incident with a ranked, explainable root cause and a bound runbook.
Vendor adapters (O1, gNMI, srsRAN) feed one message bus; a normalizer maps every vendor's format to one model; a topology-aware correlator folds related alarms into a single fault group and ranks the probable root cause with its reasons attached. One alert per incident — not 96 — and a tamper-evident audit trail behind it all.
Read the full technical brief →Every vendor update is tested against your specific multi-vendor mix before it ever reaches production.
A vendor drops new DU firmware; before it touches live traffic, the interop runner replays it against your exact vendor combination and produces a signed pass/fail verdict. Bad updates stop in the lab, and you hold neutral evidence for every SLA conversation.
Read the full technical brief →Idle accelerated edge compute becomes billable AI capacity — with a hard guarantee that radio always pre-empts.
The compute broker meters and sells the accelerator hours your radio isn't using, with metered billing and a non-negotiable rule enforced in code: radio workloads pre-empt tenant workloads, always. New revenue from hardware you already own.
Read the full technical brief →An in-country inference appliance you white-label and resell to banks, fintechs and government.
A hardened inference appliance runs inside your network under your brand: banks, fintechs and public agencies get low-latency AI whose data never crosses the border, and you get the recurring revenue with attestation and metering built in.
Read the full technical brief →Not a mockup — these are screens from the MODERAN Console during a live fault-injection demo: one dead unit, 96 alarms, one named root cause.
The boundary is the architecture, enforced in code.
Self-hosted in your own network. Your performance data, faults, audit trail and any subscriber data never leave your country. ARGILETTE operates and supports the platform but is not a processor of your data — the only thing we ever see is coarse system health (is a service up, which version). Aligned with NDPA (Nigeria), CDP (Senegal), ARTCI (Côte d'Ivoire) and the ECOWAS Supplementary Act.
A pilot is deliberately low-risk: MODERAN only observes your network — it never controls it. Your own team installs it, following the step-by-step guide, usually with no ARGILETTE engineer on site.
Deploy the platform onto your Kubernetes cluster, in your own data centre. Your staff sign in with their existing corporate accounts.
Point the collectors at the equipment you already run — O1, gNMI, srsRAN/OAI. Every vendor's data lands in one common model.
Your NOC sees its first correlated incident: one card, ranked root cause, runbook attached — instead of a hundred raw alarms.
The platform runs in your cluster and your team operates it. ARGILETTE ships signed releases, support and the roadmap.
You provide the cluster and RAN access; ARGILETTE operates the deployment under contract — physically inside your infrastructure, so your data still never leaves.
Adoption is a program, not a hand-off. An eight-module training curriculum carries your teams from installation to NOC operations — including a module delivered directly to your bank and enterprise clients' compliance teams.
Two front doors: bring the full platform to your network with a low-risk 30-day pilot, or rent sovereign AI capacity today.
Tell us about your network and we'll come back with a pilot plan — observability first, on your servers, your data never leaving the country.
Rent sovereign, in-country AI inference capacity. Self-service signup, pay by the plan, data stays in your region.
Go to AI Capacity →