Ftrain, by Paul Ford

Sure, AI might bring on the end times—but you can use it to build your own tax software! On this week’s podcast, Paul and Rich are joined in the studio by Craig Mod, a writer, photographer, and self-described “software supertaster” who recently took to Claude to build a version of Quicken that suited his complicated tax-filing needs. First, they discuss the project and assess how likely it is that a non-technologist would be able to create something similar with the current tools. Then, the conversation takes a turn—towards the apocalypse. (It’s fun, we promise!)

Doctors might be using AI to cut down on paperwork, but can these tools really be employed in clinical settings? On this week’s podcast, Paul and Rich sit down in the studio with Dr. Kamal Menghrajani, a practicing oncologist and lecturer at Harvard Medical School who was previously a member of the Biden White House’s “Cancer Moonshot” team. After she explains how AI is helping in her work, she lays out its real limitations—and discusses how these technologies can distract from more systemic approaches like better patient prevention and screening. Plus: Are her medical students allowed to use LLMs?

Why are so few engineering teams reaping the benefits of AI? On this week’s episode, Paul presents Rich with the findings from a recent report from CircleCI and Thoughtworks on the productivity of enterprise teams using LLMs. While there’s been a dramatic increase in throughput—the amount of code produced—across the board, just 5% of orgs are seeing real gains from these tools, while the majority struggle with errors, bugs, and lower productivity than before AI was introduced. As Paul puts it: “The advantages of this technology are not equally distributed.”

What’s wrong with Sam Altman? Ask the guy who spent 18 months reporting on him. On this week’s podcast, Paul and Rich are joined by New Yorker staff writer Andrew Marantz, who recently put out a lengthy article on the OpenAI chief that he co-reported with Ronan Farrow. After they dive into some of the specific details of the piece, they discuss the broader questions Altman’s position in the industry raises. If AI really is as powerful as he claims it will someday be, why are we allowing one person to have that much power over it?

What does it take to make a really good product with AI tools? On this week’s podcast, Paul walks Rich through his recent adventures building a robust aggregated newsletter tool—first to track the AI industry, then generalized and customizable for any industry. Vibe-coding platforms continue to evolve, but you still need a lot of technical knowledge to make something that really works. Is that high bar likely to lower in the coming months and years?

“The blind vibe-coding revolution is upon us.” On this week’s podcast, Paul and Rich are joined by Andrew Leland, the author of the Pulitzer-finalist The Country of the Blind: A Memoir at the End of Sight, to discuss how blind and low-vision people are using AI tools to create and adapt software to suit their accessibility needs. With limits to what any out-of-the-box software or device might do, is AI the way to give disabled people technological solutions that really work? Plus: Rich bravely makes it through the whole recording despite being surrounded by two massive Emacs stans.

Employees at the big consulting firms are being told to use AI. Are they going to automate themselves out of a job? On this week’s podcast, Paul and Rich discuss a recent set of directives from the CEO of PricewaterhouseCoopers about AI adoption and consider what they frame as the “AI consulting paradox”: To show value, consultants need to introduce AI into their client relationships, potentially cutting their own value as a result. How will consulting manage this transition in the short or long term? Plus: Paul reports on his recent conversations with a New School journalism class that visited the office and shared their feelings about using AI.

Sure, you can build “stuff” with AI, but is anyone paying attention to product these days? On this week’s episode, Paul and Rich sit down with someone who is: Hilary Mason, CEO of the immersive online roleplaying game company Hidden Door. After discussing Hilary’s background in data science and machine learning, Paul takes a spin through a game scenario (Brooklyn vampires in a fantasy tavern!) while Hilary outlines the product decisions around Hidden Door’s game mechanics, AI-related or otherwise. Plus: Rich outlines his own attempts to make a game with AI for a family gathering, which, in his words, “led to children crying.”

AI is blurring—and even destroying—the distinctions between disciplines. Do we need a new way to talk about work? On this week’s episode, Paul tests out a few of his AI-era neologisms on a skeptical Rich: Perhaps you are a “custolient,” looking to purchase the services of a “praygency” for your next project? (Yes, Paul insists the “y” in “praygency” is vital.) Are these new blended terms helpful, or just a way of talking around a very uncertain landscape?

Is the future of work sitting back and watching your company of bots plan their offsite? On this week’s episode, Paul is joined in the studio by journalist Evan Ratliff, the host and creator of the wildly popular Shell Game podcast, which is about, per the show’s description, “how Evan tried to build a real startup, run by fake people.” Evan’s AI agents were an exercise in immersive journalism (and yes, they did in fact go wild planning their offsite), but would he ever consider running a company of bots for real?