Pivot

Pivot is a platform for parents of young footballers in Australia: a weekly paper called the Tribune, an assistant that knows the family on the web and in WhatsApp, and calls with experts. Its articles have to rest on evidence and sound like one editor, and the team had no developers.

I joined in January 2026 as chief editor and content automation lead and built the platform end to end: brand, front end, back end, the CMS, the AI newsroom, a worker on a server and the deployment. The screens below are from the live site, signed in as a parent.

My part

Brought in as chief editor and content automation lead. Built the whole platform end to end: brand, product, design, web development, the AI newsroom, the CMS and the infrastructure.

Visit the site: pivot.football↗

In numbers

280

books and papers in the research library

8

steps from brief to publishing, each with the editor's approval

9

cover candidates from one image generation

19

article templates on one shared frame

What I built

01

AI Newsroom in Your Editor's Voice

From idea to a laid-out article with cover and infographic, in the chief editor's voice, with sources. A person approves every step.

The problem

A small newsroom without developers needs evidence-based articles in one voice. AI sounds machine-made, invents numbers, and the editor fixes the same things again and again.

How it works

• Brief, research, outline, draft, cover, infographic, layout, publish. Approval after each step; the research, outline and draft take edits by text or voice.
• Research puts the client's own library of 280 books and papers ahead of the web, and cites its sources.
• The editor's line edits became rules the model reads at every step. They live in the admin, not in code. Each draft gets a voice score from 1 to 10.
• One image generation gives a 3×3 grid of nine covers. The editor picks a cell and it gets upscaled.
• Its own CMS: revisions, scheduled publishing, a paywall in four levels, reading depth per article, and the cost of the newsroom by step and model.

Status

• Built: finished, not in daily use yetDeployed on pivot.football.

Also fits

Club and academy media, federations, sports medicine, regulated B2B content.

02

Primary Sources and RAG

AI that looks things up in your own library before it answers, weighs how reliable each source is, and shows where it looked.

The problem

Models invent numbers, names and quotes with confidence. For a research newsroom, a club archive or a company knowledge base, that is not acceptable.

How it works

• The client's research library: 285 books and papers chosen, 280 read into the system as 9,636 passages. Embedding all of it cost about $0.12.
• Research for an article puts that library ahead of the web, and a Spanish source can feed an English article.
• The sources come back with the draft, in a section of sources and citations at the end.
• The parents' assistant answers from the paper's published articles and the family's profile: the child's age, club and level, and what worries the parent.
• On the web it remembers past chats; in WhatsApp it keeps the last 40 messages.

Status

• Built: finished, not in daily use yet280 documents, embedded for about $0.12.

Also fits

Coaching methodology libraries, medical teams, club archives with a legend who knows only his era, law firms.

03

Design

Brand, design system, data on screen, reports and sites. Variants side by side on real content; the winner becomes the standard.

The problem

An interface made of template cards and default fonts looks like every dashboard. When every screen is solved differently, people relearn the product on every page.

How it works

• The brand and a design system in two themes, light and dark.
• 19 article templates on one shared frame, 20 interactive elements and eight topic seals, so every piece reads as the same paper.
• Five typefaces cut to two. Newsreader, with its optical sizes, holds both 10 px labels and 22 px leads.
• 13 layouts of one reference article drawn side by side before the template was fixed.
• Designed and built end to end by me, with AI tools.

Status

• Live: running in productionLive on pivot.football.

Also fits

Brand systems for clubs, agencies and media. Data-heavy dashboards. Reports that look the same on screen and in PDF. Cinematic launch sites.

Also inside

• Built: finished, not in daily use yet

  The cost of every article

  Every AI call is logged with its article, step and model, at the price of the day it ran. The newsroom sees what an article costs: the typical piece, the expensive one, and which step and model the money went to.

• Built: finished, not in daily use yet

  Expert calls by proposal

  A parent asks, an expert offers up to five times, the parent confirms one and the rest are cancelled. Each slot can be taken only once, and the desk sees Sydney and Madrid time side by side.

• Built: finished, not in daily use yet

  The assistant in WhatsApp

  The same assistant answers in WhatsApp, voice notes included. When a person from the desk steps in, the AI steps back.