We designAgentic Pipelinesfor enterpriseWordPress teams.

We train the engineers you have,
or become the ones you don’t.

AI tools changed how developers work.

system thinking // VOL 01

91% of engineering teams have adopted AI coding tools. Organizational productivity has improved by roughly 10%

(DX, 121,000 developers across 450+ companies)

Most AI tools are focused on increasing the productivity of one individual. Nothing connects that work at the team level.

There’s no integration into the workflows and processes your team uses, or if there is, they feel bolted on at the end. There’s no easy way for the context that one developer figures out to reach the next developer who hits the same problem.

You onboarded your developers.

You didn’t onboard your AI.

Every organization invests months onboarding human developers. Helping them understand the themes, plugins, and architecture so they understand the context behind what they are working on. Then the same organization gives an AI agent a blank context window and expects useful output. For AI to be successful, it needs the same thing, someone to teach it how your team works.

A pipeline changes how your team works.

Only ~6% of organizations are seeing financial returns from AI


They were ~3× more likely to have redesigned how their teams work

— McKinsey, State of AI 2025

Your deploys run on a CI/CD pipeline.
Your AI work doesn’t… yet

Shipping used to depend on who was doing it. CI/CD changed that: the same build and the same review gates on every commit, no matter who pushed.

AI-based work hasn’t made that transition for most teams. Many developers drive it by hand, with their own prompts, their own context, their own judgment.

An Agentic Pipeline is the same move, applied to AI work. The same checks and the same reviews on every piece of work, no matter who ran it. What your best engineer would catch, the system catches.

agentic pipelines // VOL 02

Inside an Agentic Pipeline

An Agentic Pipeline isn’t a tool or a platform you can buy off the shelf. It’s infrastructure and processes custom built to support your team’s work from intent to shipped code, designed around how you work with AI.

What that infrastructure looks like varies by team. Yours will fit your company structure, codebase, stakeholders, clients, and standards. What’s underneath doesn’t vary, because each part removes a failure we kept finding in teams that adopted the tools without the system: agents working blind to what your senior engineers know, AI mistakes waiting for a human to catch them, business intent degrading through every handoff, and lessons from one sprint never reaching the next.

Read about the pillars of Organizational AI →

Agentic Pipelines do the work before you ask them to

Imagine an agent in your pipeline that watches Slack for discussions about changes to a feature request and notices those changes never made it into Jira.

If the ticket still hasn’t been updated by end of day, it comments on the ticket, points to the conversation, and explains the difference between what the ticket says and what the team actually decided.


The goal is that you don’t need to ask it to do things. It knows how you work, and it gets as much of the work as it can ready for you to review.

No more manually updating documentation after a process change. The pipeline will have been watching it, and the suggested edits will be sitting there, ready for your approval.

The Requirements Watcher agent in action

engagement // VOL 03

What working with Principal WP looks like

the team // VOL 04

We’ve scaled WordPress to billions of pageviews and billions of dollars in transactions.

Stéphane Boisvert

Stéphane Boisvert

Twenty years on mission-critical WordPress. At WordPress.com VIP he reviewed the infrastructure behind some of the platform’s biggest names, and as CTO at XWP he led a team that grew to seventy engineers. Earlier he built the 600+ site multisite behind a national political party that moved millions in donations. At Principal WP, he maps where an agentic pipeline fits and what building one takes.

Mikey Arce

Mikey Arce

Mikey is a member of the WordPress Core AI team, working on how AI fits into WordPress itself. Before Principal WP he was at Automattic on WooCommerce, the software behind close to a third of the world’s online stores, and worked on some of the world’s largest publishing sites at WordPress.com VIP and XWP. At Principal WP he builds the agentic pipelines the practice runs on, the ones built with clients.

Between us, more than 30 years in WordPress, on platforms serving billions of pageviews a month and moving billions of dollars in transactions for hundreds of the web’s largest sites.

“He will go out of his way to understand ‘what problem we are trying to solve’ instead of simply deploying something that works. Quite often, we will get a suggestion we never thought of.”

— David Parsons, Senior Software Engineer, USA TODAY Sports Media Group

“He has never provided advice that we have turned away, making him a critical member of the team on the whole.”

— Mark Parolisi, Director of Engineering, MailOnline

Some of the teams we’ve worked with over the years

Microsoft
WooCommerce
Time
Vox
The Verge
Wired
CNN
NBC
Deadline
CBS Radio
Variety
Quartz
Capgemini
TechCrunch
New York Post
The Sun
USA Today
MailOnline

Every engagement starts with a conversation.

Tell us about your team and what you’re trying to solve.
No sales pitch. No proposal template.
Just a direct conversation with the person who would do the work.

    Not on WordPress?

    The thinking transfers. We design Agentic Pipelines beyond WordPress too, for sales teams, operations teams, and other workflow-heavy groups that want one system instead of fifty individual habits.

    Notes from the build

    Occasional notes on Agentic Pipelines and organizational AI in practice: what we’re building for enterprise WordPress teams, and what we’re learning while we build it.