When another publisher or reporter asks me how to get started with AI, this is the page I point them to. It’s short on purpose — these are the things I actually use or recommend, not a directory.
If you only do one thing from this page, do this one: write a context document for your organization. It is the single change that has done the most for my own work, and I wrote up how and why.
A bias in this list, stated up front: most of these people build things. The AI conversation in journalism is mostly led by people who don’t, and it shows — a great deal written about what these tools mean, very little about what they actually do. Developers and makers will tell you where the seams are, and that’s the part you need before you can judge what anyone is selling you. Go find some and listen to them.
Learning the tools
Anthropic Academy — free courses with certificates, including tracks written for small businesses and nonprofits rather than developers. This is where I send people first.
Claude Cowork — Claude working directly on your files, where you approve each step. No terminal, no code. The right on-ramp if you want AI doing real work before you want to learn anything technical.
Context Engineering Clearly Explained (Tina Huang) — twenty minutes on why what you give the model matters more than how you phrase the ask. The technical version of the point I keep making. Her channel is worth the subscribe.
Full Claude Code Tutorial for Non-Technical Beginners (Futurepedia) — a step-by-step walkthrough of building something real with Claude Code, made for people who don’t write code. The on-ramp if “AI as a coding agent” sounds like something you’d need a developer for.
AI for local news
Joe Amditis — practical AI guides written at one-person-newsroom scale, on his own site and at the Center for Cooperative Media. The closest thing to a manual for shops like mine.
LION Publishers — the independent local news association. Peer cohorts, sustainability programs, and AI training that assumes your whole staff might be one person.
Local Media Association — AI resources — case studies from publishers actually running this stuff, not vendors selling it.
OpenNews Source and SRCCON — the community where newsroom technology gets figured out in the open.
AI and the Future of News (Reuters Institute) — where the research on what audiences actually think about AI in news lives. Useful when someone asserts what readers want.
Worth your inbox
Pluralistic (Cory Doctorow) — daily writing on how platforms turn against the people who depend on them; he coined “enshittification.” He is also the most reasonable person I’ve read on AI. His new book, The Reverse Centaur’s Guide to Life After AI (June 2026), makes the distinction that matters: a centaur uses the machine, a reverse centaur is conscripted into serving it — and nothing about the technology decides which one you become. Neither hype nor doom.
One Useful Thing (Ethan Mollick) — the most consistently practical writing on working with AI, from a Wharton professor who actually tests things.
Newsroom Robots (Nikita Roy) — the podcast on AI in the news industry.
AI as Normal Technology (Arvind Narayanan and Sayash Kapoor) — the Princeton researchers behind AI Snake Oil, treating AI as significant but ordinary technology rather than as a coming superintelligence. Neither boosterish nor doom-y, which is rarer than it should be.
How I AI — practitioners walking through the workflows they actually run, step by step.
Things I should be reading too
Everything above is what I actually use. This is the honest part of the page: what I’ve been told to read, or keep meaning to, and haven’t finished. It’s here so I actually do it.
Simon Willison — near-daily writing about what these models can actually do, from inside the build. He co-created Django and builds Datasette, which plenty of newsrooms publish data with. If my line about listening to makers means anything, he is who I mean by it.
Nieman Lab — the Nieman Foundation’s daily report on the news business. Free, and where developments in AI and newsrooms tend to surface first.
The Claude Code changelog — everything else on this page teaches a concept. This one tells you what the tool can do this month. It’s terse and written for developers, but it’s the honest answer when you’re wondering whether something you read six months ago is still true.
JournalismAI (LSE) — free training and reports, if you’d rather take a course than follow a blog.
My own tools and writing
The software I’ve built for newsrooms and note-takers lives on the Digital Tools page. Longer-form thoughts are under Writing.
Occasional notes when I ship a tool or write something — no schedule, no weekly anything.