Why Zinkworks is the Ideal Partner for Ambitious Operators Building a Fully Autonomous Network

If you run OSS for a Tier 1 operator, you already know your maturity level  on the TM Forum framework defining Autonomous Networks (AN) and your target. The scale runs from Level 0, where everything is manual, to Level 5, full autonomy. Most operators sit around Level 2 today, and the ambition across the industry is to reach Level 4 and onward to Level 5. At Zinkworks we anchor everything we build to this framework because it means we measure progress against an industry standard rather than a scorecard we invented ourselves.

TM Forum Autonomous Networks levels graphic below:

That’s the honest framing for this piece. The destination is agreed. The question  is how to deliver.

Where Zinkworks operates today

Our deployed work spans Level 2 through Level 5. Not a roadmap promise, but capabilities running on live Tier 1 networks today. This is where most operators actually get stuck: letting the network act inside a domain, then act ahead of a fault across domains.

Why Zinkworks is the ideal partner

Everything we deliver sits on the same architectural blueprint ; organised as three pillars.

• The Unified Data Core brings inventory, topology, alarms, tickets, and KPIs out of their OSS/BSS silos and into a single semantically linked store, so AI has the full picture before it acts.
• The Cognitive AI Layer is where purpose-built agents and rApps perceive network state, reason across domains, and act, governed and explainable at every step.
• The Decoupled Closed-Loop Automation layer closes the loop across RAN, Core, and Transport, TMF-aligned and multi-vendor by design, so automation scales without re-architecture.

That blueprint is what makes the rest possible. One additional component turns it into reality for operators: the people who build and deploy .

Zinkworks Expertise.

Telco OSS is our DNA and AI is our native tongue. Our people carry serious depth in both, with the large-scale Tier 1 implementations to back it up. We know what to do because we have done it before, on networks that look like yours. We also know how to embed the work in your organisation so that the value lands in operations and turns into concrete business outcomes.

Those are claims. Here’s the evidence.

“Autonomous networks are not a destination operators reach on their own. They’re built level by level, with partners who’ve already deployed in production at scale on a live network. That’s the role we’ve taken with ISPs such as Vodafone, Virgin Media O2, and MasOrange;  it is the one we’ll take with the ambitious operators wanting more value out of their network.” said Paul Madden, CEO, Zinkworks.

Proof in production, across multiple operators

These are live deployments, not lab demos or slideware.

Virgin Media O2, AI automation. We’ve had AI-driven automated network management running in production on VMO2’s fixed broadband network for more than two years. Repair times dropped by more than a third, and engineer site visits fell by 12% across fixed broadband operations. The results held across seasons, traffic spikes, and infrastructure changes, which is why VMO2 has since expanded the work to mobile RAN, core systems, and network operations.

Vodafone Network Innovation. We partnered to prove that deployment time of Network Management rApps can be compressed from months to just a week; delivering value 10x faster with a repeatable and efficient process giving the commands back to RAN Managers.

MasOrange, digital twin. We deployed our Network Digital Twin, a continuously updated virtual replica of the live network, with GenAI powered Root Cause Analysis and Anomaly detection; so the team can simulate a change before it touches production and reduce operational risk. This is how you earn the confidence to let automation make decisions: by modelling their consequences first.

What comes next

The telecommunications industry sits at the center of a harder question: “who controls the AI stack”. As networks become dependent on AI to operate, sovereignty becomes a live operational concern, not a theoretical one.

Zinkworks works with global technology partners, including Google, to shape what Sovereign AI looks like for telecom operators. We’ve defined three core principles that underpin our Sovereign AI approach:

• Governance sets out the rules: Open frameworks such as the EU AI Act, ETSI TS 104 223 and others define what “safe and ethical” means in practice.
• Security defends the systems: Robust controls protect AI models, data and infrastructure from threats and unauthorized access.
• Sovereignty makes them enforceable: Locally controlled infrastructure, data & models give CSPs the operational power to apply their own governance and security frameworks.

Zinkworks balance control and collaboration when operators want to leverage leading AI automation technologies.

First, we manage interdependence by mapping the AI stack and deciding which layers must be sovereign and which can be partnered. Then, federated learning allows local model training on sovereign data, so customers gain global intelligence without exposing sensitive data. Finally, Zinkworks’ guardrails and intent orchestration profiles every LLM-generated command to block unsafe actions before ever reaching the network.

Let’s talk at DTW

Autonomous networks get built level by level, in production, on real operator infrastructure. Everything in this piece is work we’re already doing with Tier 1 operators around the world today.

If that’s the kind of partner you’re looking for, let’s talk and come find us at DTW. We’ll be running live demonstrations of many of our AI powered solutions like Energy Saver, Digital Twin, Root Cause Analysis, Anomaly Detection and our rApp portfolio. We will also have several of our executives on stage across the week.

Stop by the stand to talk through your current architecture, your autonomy targets, and where you’re stuck. Bring your hardest question. We’ll bring the engineers who can answer it. We look forward to partnering with the most ambitious operators in building fully autonomous networks.