Documentation, AI agents and the future of tech writing.
I'd Rather Be Writing Podcast follows technical writers, academics and documentation practitioners as they work out what AI means for the job. Conversations about agent skill files sit beside recorded conference talks and an AI book club.
PRESS PLAY
I'd Rather Be Writing Podcast is a running conversation about technical writing now that AI sits in the middle of it. Tom Johnson turns up in each recording here, whether he is leading a discussion, giving a talk or sitting in with a book club. One episode opens by naming the subject as the usual one: the intersection of documentation and AI. The feed does not stick to one format. Some episodes are small-group conversations with a guest, and Fabrizio Ferri-Benedetti of Paso Uno takes part in some of them. Others are recorded talks, including a MEGAComm conference session and a campus presentation that pauses for audience questions. There is also an AI Book Club, which meets on a Sunday morning and is recorded for people who cannot join. The recording here covers 'If Anyone Builds It, Everyone Dies'. The practical episodes get specific. Manny Silva, who is writing a book on docs as tests, defines a skill as a markdown file that tells an LLM how to perform a task. He counts about 140 of his own, and some are whole workflows that assign tasks to different agents. The conversation with Larah Vasquez, who works for Tailscale, stops to explain git worktrees: local copies of a repo that let several branches, and several agents, run side by side. Other episodes step back from the tools. The campus talk on cyborg technical writers argues that the job now means less writing and more coaching, steering and correcting AI, with writing becoming a medium rather than the skill itself. Its closing exchange asks whether companies spending billions on AI will ever see a matching return. Two professors, Nupoor Ranade and Jeremy Merritt, bring the academic side, including how scholars have kept redefining the field since around the mid-1960s. That conversation ends with a wish for better communication between practitioners and academics. The MEGAComm talk is more personal: a tech writer with an audience online can still feel like a mouse on the wall inside the company. The book club goes darkest, as participants recount reports of AI blackmail and of a chatbot encouraging a teenager toward suicide. Several titles are phrased as open questions, such as whether tech writers will survive AI and how valuable agent skills are. For hands-on agent workflow, start with the Manny Silva and Larah Vasquez conversations. For the argument about where the profession is going, start with the professors' episode or the cyborg talk.
Made for: Suits a technical writer, documentation lead or content strategist who wants practitioner-level talk about skill files, agent workflows and what AI means for the job. It also suits someone curious how academics in technical communication are reading the same changes.
What sets it apart: It puts detailed talk about skill files and git worktrees in the same feed as a conference session, a campus talk on cyborg technical writers and a book club on AI risk. Tom Johnson is the common thread across all of them.
Topics and issues for technical writers. Improve your technical communication role. Empower users with better documentation.
An AI agent needs babysitting, and Tom Johnson wants to know what that makes your writing job.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A practical look at agent workflows, work trees, Git, and the changing mechanics of documentation.
Agent skills and workflow habits get a grounded look as the conversation moves into the day-to-day mechanics of documentation. The focus stays practical: specialized operations in a separate context window, work trees that keep branches isolated, and the ways an LLM can take on…
AI pushes technical writing toward directing the work around the words.
Technical writing comes through here as systems work, not just wordcraft. The cyborg model puts the writer in a hybrid role: bringing judgment, training, and context to AI, then spending much of the job coaching, steering, correcting, and repairing what comes back.
Two assistant professors place generative AI within technical communication's long pattern of adaptation.
Generative AI enters the episode as a question about the future of technical writing, but the conversation is strongest when it treats that future as part of a longer pattern of change.
A book-club debate ties AI catastrophe to blackmail, machine desire, and the writing people still want to keep human.
The AI Book Club takes a doomy book about AI risk and turns it into a practical conversation about whether artificial intelligence can form intentions of its own.
Manny Silva explains how skill files can turn documentation habits into usable instructions for AI agents.
Documentation gets a grounded, useful treatment as Tom Johnson and Fabrizio Ferri Benedetti talk with Manny Silva about where documentation and AI meet.
Clearer docs, sharper systems.
Specialists walking through software's tradeoffs, one system at a time.
Founders and engineers on AI coding agents like Claude Code, and supply chain security.
Tom Johnson interviews Claudio Descalescu about Archby, a new documentation tool for developers.
Tom Johnson discusses his experience presenting to a Chinese audience and fills in for another speaker at a technical writing event.
A roundtable discussion among technical writers about how AI affects the writing profession.
Tom Johnson discusses the paradox of being well-known online but feeling invisible within his own company as a tech writer.
Tom Johnson, Fabrizio Ferri Benedetti, and Lara Vasquez spend much of the episode on AI tooling for documentation, with Lara tying her Tailscale work to Write the Docs and GitHub.
A technical writer discusses current trends and strategies for success in the field.
Tom Johnson talks with Jeremy Roslett Merritt of James Madison University and Nupur Ranade of Carnegie Mellon University about whether generative AI threatens technical writing.
The AI Book Club discusses Nate Soares and Eliezer Yudkowsky’s If Anyone Builds It, Everyone Dies, a book the group repeatedly calls controversial and centered on AI safety and alignment.
Tom Johnson, Fabrizio Ferri Benedetti, and Manny Silva discuss documentation and AI, with Manny positioning DocDetective and his book Docs as Tests around deterministic checks for procedures, CLI steps, screenshots, and API commands.
Four tech workers discuss Sarah Wynn-Williams’s Careless People, weighing its account of Facebook against questions about complicity and hindsight.
Tom Johnson and Fabrizio Ferdi Benedetti compare 2026 predictions about AI, regulation, and technical writing.
Tom Johnson’s AI book club discusses Megan O’Giblin’s God, Human, Animal, Machine and its parallels between religion and artificial intelligence.
Tom Johnson discusses the paradox of being well-known online but feeling invisible within his own company as a technical writer.