Agile Telco: What are the key attributes of an autonomous telco?

Fari Pirouz: An autonomous telco has three defining attributes. First, it orchestrates decisions end-to-end, not in silos. One order, one journey, one source of truth. Second, it treats autonomy as an operating model, not a technology layer. The centre of gravity moves from building intelligence to operating it. Third, it builds reusable patterns. Autonomy scales through re-use rather than reinvention. An autonomous telco captures how decisions were made in the first domain and inherits that logic in the next.

The architecture is model driven and centre-out by having a shared core of decisioning that radiates into every channel and domain. Finally, an autonomous telco is transparent. Decisions happen inside governed workflows with full auditability. The question that matters is whether the organisation can trust that decision in production.

Agile Telco: How are telcos adding autonomy across operations, not just at networks?

FP: Telcos are adding autonomy everywhere from fault detection, capacity planning, service provisioning, billing workflows and customer support. But they’re doing it domain-by-domain with no orchestration connecting them, creating pockets of intelligent automation that don’t know what others are doing.

The operators scaling autonomy beyond networks take a different approach. They commit to governance, auditability and traceability as the foundation before picking any tool. A telco decides which operations need autonomy and, before selecting platforms, defines a shared decision model. Define the decision model before automating by clarifying three things: what is the desired outcome, what inputs drive the decision, and what policies or constraints must be enforced?

Agile Telco: What are the challenges of enabling autonomous operations for telcos as regulated providers of critical infrastructure?

FP: Regulating critical infrastructure adds complexity to autonomous operations. When a network fails and regulators investigate, they ask for logs. A log file shows what happened, not why. It records that a system took an action but doesn’t explain the decision logic behind that action.

For autonomous operations in regulated telco infrastructure, this gap is fatal. The answer is full decision level traceability across all systems whether the decision was made by an autonomous agent or a human operator. The traceability also needs to answer three key things from what inputs were considered, what policy was applied and why was this decision chosen over alternatives?

Telcos treating this as a compliance checkbox after deployment will struggle. Those building decision level traceability into the architecture from day one, where every autonomous action creates an auditable record of reasoning, not just outcomes, will operate with regulatory confidence.

Agile Telco: How will telcos integrate islands of autonomy to create fully-autonomous systems and processes?

FP: Most telcos have autonomy pockets everywhere. The natural instinct is to connect them with point-to-point plumbing, such as through API-to-API. That’s just integration not orchestration, in reality it’s just gluing islands together.

Operators making real progress ask, what do truly autonomous telco operations look like end-to-end? They pick high impact, repeatable use cases like fault resolution, capacity optimisation and service provisioning. They don’t just automate that domain but design how it will connect to adjacent domains.

You must take a more strategic approach. Before picking tools, before selecting vendors, before launching pilots, I think you need to answer how does this autonomous capability connect to the next one? What governance must be shared? What decision logic must be consistent? You should only start small on an autonomous initiative when you understand how it scales. Otherwise, you’re just building for failure.

Agile Telco: What does a good strategy to achieve autonomous operations across a whole telco look like?

FP: Most telco strategies for autonomy are really just launch plans for a pilot. Pick a domain, build an agent, measure return on investment (ROI), scale if it works. It almost always fails because pilots don’t represent production.

A real strategy starts with value, not technology. What’s the value you want from autonomy in terms of ROI and total cost of ownership (TCO)? This shifts from ‘can we automate this?’ to ‘should we automate this?’

TCO depends on how your autonomy platform prices itself. If paying per token or per decision, TCO scales with success. A fixed cost autonomy model eliminates that constraint. Your cost structure is flat. As you scale, per unit costs decline.

Not everything should be autonomous. Some decisions benefit from human judgment. Some processes are so tangled with legacy systems that autonomy becomes expensive. A good strategy should ask where autonomy plays a role and where it doesn’t. Most real operations will be hybrid, with autonomy handling routine decisions and humans handling judgment calls.

Agile Telco: What are the barriers and opportunities and when will we see truly autonomous telcos operating?

FP: Legacy systems are a real barrier with the messy data, fragmented integrations, decision logic scattered across old tools but that barrier is also the opportunity.
Addressing legacy systems with connected autonomous solutions creates tremendous opportunities to reduce operational expenditure. Every telco watching opex climb faces pressure to do more with less. Autonomy is one of few levers that actually works.

The opportunity is multi layered. First, obvious efficiency and routine operations can be automated. Second is the speed, operations that took weeks can now be done in hours. Greater automation increases efficiency, time to market and reduces design time on new products and services.

To achieve fully autonomous telco operations, we need to be realistic and think of a timeframe of several years. Autonomy exists in pockets today and linking all these up is a challenge that’s well worth pursuing.