Skip to content

Autonomous Networks — Level 4 and beyond

Networks that think, act, and
self-heal.

Zinkworks delivers the AI platform, agentic rApps, and deep OSS expertise that move operators from reactive operations to fully autonomous, zero-touch networks.

Book a demo
12 wks Deployment time — down from 2 years
80% Reduction in network operations tickets
92% AI model accuracy — traffic prediction
30 kWh Energy saved per base station, daily

The Gap

59% of operators are stuck in the middle.

TM Forum defines six levels of network autonomy, from manual operations at L0 to fully autonomous at L5. The chart below plots the industry against the ladder — most operators make progress in pockets and can't repeat it elsewhere, so the picture stays roughly the same.

  • L0Manual3%Pre-digitalOperations & MaintenanceAll operations are handled manually. Knowledge lives in spreadsheets and with individual engineers, and issues are often discovered only after customers report them.
  • L1Assisted11%Legacy OSSTooling, not automationMonitoring and alerting tools provide visibility, but diagnosis and resolution remain manual. Data is siloed, alarm volumes are high, and root-cause analysis is slow.
  • L2Partial59%You are hereScripts in pocketsPre-built scripts automate routine tasks, but complex, cross-domain, or new issues still require human intervention. Progress depends heavily on specialist expertise.
  • L3Conditional21%ReachableClosed loops, narrow scopeClosed-loop automation works for specific scenarios, but each new use case requires additional engineering and integration effort.
  • L4High6%Where we helpIntent-driven, predictiveIntent-driven and predictive operations enabled by a unified data layer, knowledge graph, and AI orchestration. Adoption remains limited despite strong industry ambitions (6% today, 73% by 2030).
  • L5Full<1%In progressSelf-organisingSelf-organising, self-healing networks operate autonomously across all domains. Key capabilities exist today, but fully end-to-end autonomous operations are still being proven.
3%
11%
59%
21%
6%
<1%

L2 · You are here

Partial — Scripts in pockets

Pre-built scripts automate routine tasks, but complex, cross-domain, or new issues still require human intervention. Progress depends heavily on specialist expertise.

59%

of operators
at this level

The Integration Challenge

Operators rate cross-domain integration as the single biggest obstacle — up from 53% the year before. These are engineering problems before they're AI problems.

60%

Cross-domain
integration

52%

Legacy
software

44%

Unclear
L4 path

Evidence

Active programmes and ongoing industry work.

A short summary of where the work is live, where we'll be presenting it, and the partnerships behind it. TM Forum's own data shows what good looks like at Level 4: China Mobile saving 6,000 hours of engineering work per year, DNB cutting MTTR by 54%, Telstra hitting 30% faster time to market on 5G slicing. The work below is how we help operators get there.

VMO2 Fireside Chat — Zinkworks and Google Cloud VMO2 Fireside · 2026

VMO2 · Active Engagement

Full ANO programme with VMO2

We're currently running an Autonomous Network Operations programme with Virgin Media O₂. It covers data foundation, knowledge graph, multi-agent orchestration and closed-loop automation across both fixed and mobile. The Zinkworks journey applied end-to-end at a Tier-1 European operator, including the early L5 components we're now putting into production.

Use cases live2
MaturityL2 → L4 · L5 in flight
ScopeMobile + Fixed
Built onGoogle Cloud
Partnership · Google Cloud GCP partner

Agentic AI on Vertex AI & Gemini Enterprise

Established GCP partner. We work with Google on bringing newer technologies into telco environments, including Agentic AI on Vertex AI and Gemini Enterprise. Worth noting: 63% of CSPs plan to introduce agentic AI into network operations, with only 15% having actually done so. That gap is where our work sits.

Heritage ~200 engineers · Ireland

From the Ericsson OSS Competence Centre

Zinkworks was spun out of Ericsson's OSS Competence Centre. Around 200 engineers based in Ireland, with a background in telco transformation and virtualised RAN/Core orchestration.

Catalyst TM Forum

TM Forum Catalyst contributions

We have contributed to TM Forum Catalyst projects, including Sustainable 5G: Smart RAN Energy Management: a catalyst focused on AI-driven RAN energy optimisation to help operators reduce energy consumption and improve the efficiency of 5G network operations.

Common questions

01 01 What are Autonomous Network Operations (ANO)?

A structured AI programme that automates how network operators monitor, diagnose, and fix their networks, We replace manual, reactive NOC work with intelligent, data-driven capabilities that surface insights and, eventually, take governed corrective actions automatically.

02 02 What can ANO do?

We have two user-facing surfaces: an AI Copilot and a Digital Twin of the network, underpinned by analytic engines covering areas like energy saving, firmware hygiene, anomaly detection, fault correlation & RCA and capacity planning. The specific mix is shaped by where the operator is on their journey.

03 03 Does it actually change the network, or just advise?

Operators need to build up their trust in any new software. We therefore work on a case-by-case, gradual journey where we start with an advisory model, and later evolve into full automated, closed-loop actions. The underlying engines are identical in both modes where the "Autonomous Actions" step is switched on when ready.

04 04 How do you decide where to start?

The starting point is shaped by what data the operator already has, where the pain is greatest, and what the business is trying to achieve such as cost reduction, resilience, sustainability, or compliance. Our engagements begin with discovery to map current data readiness against potential value, and the roadmap follows from that.

Book a Demo

See autonomous networks in action.

Tell us where you are on the path to Level 4 and we'll show you the agents, models and platforms that match. A 30-minute walkthrough on your network, your domains, your data.

Live RCA, anomaly detection and proactive fault prediction
Closed-loop automation across RAN, Core and Transport
Ericsson EIAP, Samsung SMO and Google Cloud integrations
Tell us what you want to see

Enter your contact info

Consent(Required)
Consent