← Selected work

AI Ops function building · Strategy · Governance · Enablement

Building AI Operations from the Work Up

How do individual AI projects turn into a company-wide operating model instead of thirty parallel pilots?

Company
SoundCloud
Period
2024–2026
Disciplines
AI Strategy, Governance, Org Design, Platform
Chapters
8
6
Functions on one roadmap
1
Enterprise access, routing, governance layer
AI Collective
Internal community of practice
01The situation

AI adoption at SoundCloud was happening the way it happens everywhere: enthusiastically, in parallel, and without a shared spine. Teams across product, engineering, legal, marketing, trust & safety, and operations were each evaluating vendors, writing their own prompts, and forming private opinions about what was safe. The energy was real. The duplication and risk were also real.

02The actual problem

The request was

Pick our AI tools and set the policy.

Discovery showed

The company did not have a tooling problem. It had a decision-making problem, no shared way to judge which work should be automated, who owned the outcome, and what evidence made a use case safe to ship.

A policy alone would have produced governance theater: a document nobody reads, applied by a committee nobody wants to schedule. What was actually missing was an operating model, access, routing, and guardrails built into the path of least resistance, plus a roadmap that ranked opportunities by impact rather than by whoever asked loudest.

03What I learned
  • Interviewed leaders across six functions to inventory in-flight AI experiments and the decisions each one was actually blocked on.
  • Mapped support, trust & safety, monetization, and internal ops workflows against volume, cost, error tolerance, and reversibility.
  • Analyzed spend and shadow usage to size the real cost curve rather than the projected one.
  • Pressure-tested legal and privacy constraints early, so the roadmap was built inside the real boundaries instead of discovering them at launch.
04How I framed it

Exhibit

Task Autonomy Framework

 Knowledge depthAI suitabilityIntegrationAutonomy level
01Tier 1: Repetitive, low ambiguityShallow, documentedHighRead-onlyFull autonomy, sampled review
02Tier 2: Judgment inside known rulesPolicy + contextHigh with retrievalRead + writeAutonomous, human on exception
03Tier 3: Judgment with real consequenceTribal + policyAssistiveWrite with auditHuman in the loop, always
04Tier 4: Novel, contested, or regulatedExpert onlyDrafting onlyNoneHuman decides, AI supports

The tiers did the political work: teams stopped debating whether AI was 'safe' in the abstract and started placing individual tasks on a shared ladder.

What you're looking atA framework I created to evaluate operational tasks across knowledge requirements, AI suitability, integration needs, and appropriate autonomy, so ‘should this be automated’ became an answerable question instead of an argument.

Exhibit

Enterprise AI Reference Architecture

01Governance & policy

Data classification, acceptable use, vendor risk, retention

02Evaluation & telemetry

Eval harnesses, quality signals, logging, cost and latency

03Orchestration & agents

Shared prompt and tool registry, agent patterns, handoff rules

04Model access & routing

One gateway, per-team budgets, model selection by task tier

05Data & retrieval

Classified sources, retrieval boundaries, audit on sensitive flows

Shared infrastructure, not point solutions, the difference between 30 pilots and one platform.

What you're looking atCentralizing access, routing, and telemetry meant every team inherited cost controls and audit trails by default rather than rebuilding them per project.
05What we built
  • 01An AI-first operating model aligning product, engineering, legal, marketing, trust & safety, and operations around one strategy.
  • 02A data-driven AI roadmap that surfaced the opportunities with the biggest payoff across the business and sequenced quick wins against platform investment.
  • 03Centralized enterprise AI access, routing, and governance with cost and risk controls built in.
  • 04The AI Collective, an internal community of practice that accelerated responsible experimentation without approval friction.
  • 05Production AI agents in customer operations, plus the guardrails, evaluation patterns, and playbooks that made them repeatable.
She consistently cut through the noise, evaluating opportunities and pressure-testing their potential.
Maria Von Watzdorf, VP, Product at SoundCloudMaria Von WatzdorfVP, Product at SoundCloudRead in full →
06My role
Strategy
Owned the company-wide AI strategy and roadmap end to end.
Architecture
Set the reference architecture with engineering and data partners.
Governance
Built classification, access, and vendor-risk standards with legal and security.
Facilitation
Founded and ran the AI Collective; office hours, demos, enablement.
Operating metrics
Defined containment, CSAT, and deflection targets; reported to executives.
07The outcome
  • Reduced ticket volume and increased self-resolution through deployed customer-operations agents.
  • Moved AI spend from untracked shadow usage into governed, budgeted, attributable routing.
  • Established containment, CSAT, and deflection as executive-visible operating metrics.
  • Left behind reusable AI Ops patterns that continued to scale past the original implementations.
08What I'd do differently

I underestimated how much of this job was social. The architecture was tractable within weeks; the trust took quarters. If I did it again I would launch the community of practice first and let the platform follow the demand it surfaced, rather than the other way around.

Contact

Let's get into it.

Ambiguous problem, AI adoption that stalled, an operating model that stopped scaling, a support function that should be a product. That is the conversation I want.