Anthony DeRosa
Editorial operations · Content systems · AI-era workflows

Anyone can publish once. I build the machine that does it every week.

Teams, workflow, standards, distribution, and measurement — assembled into an operation that holds its quality bar without me in the room.

4person editorial team hired and managed from scratch
209page WSJ subscription roadmap, co-authored and executed
10 → 13%WSJ digital subscription growth, year over year
4publishing products running on one operating model

Six parts of the machine.

Most content operations have two or three of these. The missing ones are why quality slips the moment volume goes up.

01

Team and roles

Who decides, who makes, who checks. Most quality failures are actually an undefined decision right.

Hired and ran a 4-person team at Pluralsight
02

Workflow and cadence

A pipeline with named stages and real deadlines, sized to what the team can hold every week — not on a good week.

3 newsletters + a podcast, weekly
03

The quality bar

Written standards for sourcing, expertise, and voice, so a piece can be judged against something other than taste.

Co-authored the Verification Handbook
04

Distribution and segmentation

The platform layer: who gets what, when, and why. Coverage is worthless if it lands in one undifferentiated list.

Mailchimp → Iterable migration, enterprise segmentation
05

Analytics people act on

Numbers routed to the person who can change something, in language they already use. Dashboards nobody opens are decoration.

Bureau-level analytics across the WSJ newsroom
06

AI-assisted production

Machines on the parts that are mechanical, humans on judgment and verification. The line moves every quarter and has to be re-drawn deliberately.

Eight AI systems built end-to-end
Case study · Pluralsight

No platform, no list, no team. Then a publishing operation.

Day one

A mandate to reach senior technology decision-makers, and nothing to do it with. No publishing platform, no subscriber list, no editorial team, no standards, no analytics.

What got built

Hired and managed four editors. Defined the coverage areas, cadence, and quality bar across AI, cybersecurity, cloud, and workforce transformation. Migrated Mailchimp → Iterable and built segmentation for an enterprise subscriber base. Wired the output into sales, CS, and enablement.

The state it was left in

10,000+ subscribers in under a year against a 5,000 goal, 65%+ open rates, 6%+ clicks — produced by a documented operating model a team could run without me.

Run at three very different speeds.

A nightly comedy show, a global business newsroom, and a B2B intelligence program need the same discipline and completely different metabolisms.

The Daily ShowNightly · cultural speed

Helped build a digital department that could turn a strong editorial voice into platform-native execution the same day — social, video, live experiences, books, and activations.

Emmy Award
The Wall Street JournalDaily · global newsroom

Strategic newsroom team reversing a subscription plateau: a 209-page roadmap, a flagship newsletter relaunch, and bureau strategies driven by analytics.

3.7M subscribers
PluralsightWeekly · B2B intelligence

A market intelligence operation for technical buyers, built from zero and positioned as independent coverage rather than content marketing.

10K+ in one year
The 2026 question

What AI actually changes about editorial operations.

It collapses production cost and leaves judgment as the only scarce input. That inverts the org chart of a content team: fewer people producing drafts, more people deciding what is worth saying, verifying it, and owning the standard. The operations problem becomes quality control at volume, not volume.

I’d rather show than argue. I’ve built eight AI systems end-to-end — a context layer, an ops layer, and full editorial products with CMSes and reader apps. That’s the same architecture a company needs for market intelligence, tested on my own workflows first.

See the eight builds

Publishing more
than you can hold?

Usually it’s not a headcount problem. It’s a missing workflow, an undefined quality bar, or analytics nobody acts on.

Get in touch