Operational Simplicity for the AI Era: Why I’m Joining Alkira’s Advisory Board

By Maria N. Martinez

For decades, I’ve had the privilege of leading at the intersection of customers, operations, and technology—at companies where the stakes are measured in mission-critical outcomes. Most recently as COO at Cisco and previously leading global customer success at Salesforce, my north star has been consistent: delivering real, repeatable customer value at scale.

Today, the next “scale challenge” is clear. AI is moving from pilots to production, and networks—too often treated as plumbing—are becoming the deciding layer between AI ambition and enterprise reality. That’s why I’m joining Alkira’s Advisory Board: to help an exceptional team accelerate a modern, service-based approach to global networking that is simple, deterministic, secure, and ready for AI.

What customers are telling us (and what’s needed next)

In conversations across industries, I hear the same three imperatives:

1. Simplicity with guardrails. Complexity kills time-to-value. Teams want outcomes, not assembly instructions or box-by-box rollouts. But simplicity must come with policy-safe guardrails: segmentation by default, least-privilege reachability, and explainability that auditors and executives can trust.
   
2. Deterministic performance for AI. Training runs, inference pipelines, and agentic workflows break when paths are best-effort. “Probably fine” is not a reliability strategy when the job queue is a multi-million-dollar GPU cluster and service windows are hours, not quarters.
   
3. Operational leverage. NetOps needs intelligence that proposes, simulates, and verifies change—so people approve, not implore. AI should reduce toil and variance, not introduce another black box.
   

These aren’t feature requests; they’re prerequisites for running the business.

Why Alkira

Alkira approaches networking as a cloud-delivered fabric—consumed as a service, not assembled—so enterprises can unify sites, clouds, colos, partners, and users without new hardware or controller sprawl. The promise is straightforward: design in minutes, enforce policy by default, scale elastically, and verify continuously. That alignment with customer outcomes is the reason I’m here.

Two elements stand out:

• An AI-native operational model. Alkira’s Network Infrastructure Assistant (NIA) and Model Context Protocol (MCP) server exemplify how AI should show up in operations: with transparency, simulation, and auditability. Natural-language intent is valuable only when it compiles into safe, reversible change with policy guarantees.
  
• Security inside the fabric. Zero-trust segmentation, policy-driven routing, and service-chaining shouldn’t be bolt-ons. They should travel with the workload. This “security-in-fabric” stance matches how mature enterprises govern risk—natively, consistently, and measurably.

Lessons I’m bringing with me

From Cisco and Salesforce, and from serving on the boards of companies that operate at formidable scale, a few principles have proven durable. 

1. Customer value is an operating system, not a slogan. Durable companies make outcomes measurable and visible. For networking, that means defining the 3–4 signals customers actually feel—how quickly they get connected, how predictably policies are enforced, how confidently changes are made, and how clearly operations are explained—and holding ourselves accountable to improving those quarter over quarter.
   
2. Simplicity scales, complexity compounds. Every runtime knob becomes a future incident. Architectures that minimize surface area, standardize patterns, and automate guardrails will always out-execute in the long run. Service models beat bespoke builds.
   
3. Trust is a feature. In regulated and safety-critical environments, trust accrues from explainability, traceability, and reversibility. If an AI system proposes a change, it must also simulate the blast radius, produce a diff the human can approve, and guarantee a clean rollback.
   
4. Great cultures practice consequential clarity. The best teams decide quickly, instrument relentlessly, and learn in public. They’re unafraid to zoom out to strategy and then zoom right back into the metric that pays it off.

A customer-backed definition of “AI-native networking”

“AI-native” is not a label; it’s a contract. Here’s how I define it in operational terms:

• Intent you can verify. Natural-language requests compile into structured policy and simulated network changes with provable constraints.
  
• Deterministic paths for AI-sensitive traffic. High-throughput, low-jitter routes to data lakes, GPU clusters, and edge inferencing—selected, monitored, and (when needed) re-routed by policy, not hope.
  
• Security that travels with the workload. Identity- and attribute-based segmentation, service-chained controls, and compliance artifacts generated as a byproduct of doing the right thing.
  
• Human-centered automation. The operator approves; the system explains. Every action is logged, every rollback is rehearsed, and every “why” is answerable.
  

This is the bar customers should set—and the one providers should meet.

What success looks like

In my experience, durable success shows up as repeatable outcomes:

• Time-to-value: Days to first secure segment and policy-verified connectivity across clouds and sites—measured, not marketed.
  
• Operational efficiency: Tickets per 1,000 endpoints drop and changes per operator rise—without safety tradeoffs.
  
• Risk reduction: Detect policy drift before it matters and generate compliance evidence continuously, not quarterly.
  
• Business agility: New AI workloads and external partners connected in hours, with deterministic SLAs and predictable cost envelopes.

If we can’t quantify these, we haven’t earned the word “platform.”

My commitment

I’m joining Alkira to help the team codify customer value into the operating fabric: how products are designed, how outcomes are measured, how changes are explained and approved, and how trust is earned release after release. I’ll bring the customer’s perspective to every decision: does this reduce time-to-value, does it improve safety and predictability, and does it scale with less effort tomorrow than it required today?

Enterprises don’t need more components. They need confidence. The network must accelerate AI, not constrain it. Security must be built in, not bolted on. Operations must be explainable, not opaque. 

That is the standard I will drive, and I’m excited to join Alkira to build a better networking future together.

Background: Maria N. Martinez is a former EVP & Chief Operating Officer of Cisco who previously led global customer success at Salesforce and now serves on the boards of McKesson, Tyson Foods, and Bank of America.

Announcement reference: “Former Cisco Systems COO Maria Martinez Joins Alkira’s Advisory Board.”