Streamline release notes: balance AI and human oversight

Tashfeen Mahmud, senior technical writer at Samsung Research in Canada explains why curiosity is your best defense against technical aversion, and how to prepare for a future where technical writers are primarily AI trainers. 'Adapting to AI isn't like learning to cook from scratch. It’s like walking into a kitchen that already knows how to cook: it just needs a human to guide it through the process. Your new role is not to execute every step, but to orchestrate an AI that already possesses the technical capability.' '[...]that takes you to decide which AI solution should you even use? If you're a large company, it's likely you will want to build your own solution or you'll get into a very expensive, almost exclusive contract with some AI provider, let's say on ChatGPT Enterprise or something like that. But if you are a small team, then that means you will probably gravitate towards a hosted solution, like a lower tier chatGPT account or something like that, or Claude or something else. Or if you're kind of picky about security and control, you will probably try hosting your own smaller open source model and training that, basically getting it to be workable for your purposes.'

Senior technical writers Gareth Brinn and Lydia Pedersen talk about using GPTs and n8n to automate release notes: scripting, prompt engineering, and synthesizing data from Jira and GitHub. Next plans include using AI for first drafts of core docs while maintaining oversight. Their advice: start simple, iterate, and grow to achieve process improvements and strategic benefits.

API thought leader Kevin Swiber and Pronovix CEO Kristof Van Tomme join the podcast to explore the Model Context Protocol (MCP) and the emerging governance challenges it presents to API strategy. We discuss the need to balance driving business value and opportunity against mitigating the security risks associated with rapid MCP adoption, noting the pressure this technology puts on enterprises to safely enable its use. MCP and Large Language Models (LLMs) are changing paradigms from simple access control (previously focused on preventing access to the wrong information) to managing and bridging context between different systems, such as a workflow for example between  a corporate context and a remote context.

We discuss findings from GitBook's "State of the Docs" report (2025), for example that 90% view documentation as crucial for purchases, yet 40% of those interviewed don't track its performance. We talk of speeding up API integration, and GitBook's focus on refining adaptive and personalized documentation features, then address avoiding tunnel vision and keeping discoverability for users, underscoring the ongoing innovation in the field. Bio: Addison Schultz is a web developer & designer working and living in Tokyo, Japan. He leads the developer relations team at GitBook, a tool for technical teams to document their products and APIs. He's been working the last 7 years on collaborative tooling, worked with some of the world’s leading design and development companies, providing product teams with advice and industry advice on how teams can build better products, faster.References: https://www.stateofdocs.com/2025/https://gitbook.com/docs/https://apithedocs.org/ai-docs-online-2025

Welcome back to the A(P)I The Docs podcast! We’re opening this new chapter with an absolute treat of an episode in which Anila Zaidi, Staff Technical Writer at Squarespace, and Marco Spinello, Senior Technical Writer at Booking.com discuss the transformative role of generative AI in technical writing and beyond. We explore how AI has evolved from an experimental tool to an essential aid whilst also touching on the challenges and ethical considerations surrounding AI.

How do you get people on board when you shift to a new tool or a new approach? Even if the value proposition is clear, it might be a challenge to create a practice out of the change.  In this episode of API The Docs, Polina Zaichkina (Senior technical writer at Codat) and Max Clayton Cowes (Product director for experience at Codat) share insights about building a culture of participation around documentation and achieving internal enablement.

In this episode, Naomi Pentrel and Andrew Feierabend share their journey of revamping Viam’s robotics APIs documentation. They discuss the benefits of sitting down with actual users in well-structured interview sessions with specific scripts and tasks. Learn about their approach to user research, their team’s collaborative methodology for achieving daily usability wins, mentoring, and implementing larger-scale changes like automation. And it all started with a smart cat feeder!

Byrne Reese had the opportunity to build a number of developer programs over the years. In this episode, he explains his unique Developer Pyramid framework which he uses as a guide to build a successful and scalable developer program now at RingCentral. He also dives into his passion for board games and how it influences his approach to documentation, and explores the Workflow Builder tool that redefines building on the RingCentral platform. Tune in for all the details!

When it comes to learning Git, it can be daunting to navigate the plethora of learning resources out there. Recognising the need for a visually engaging all-levels experience, Anna Skoulikari (JetBrains) decided to write her own self-study book to help learners along the way.

"Key results are the things you're going to do to move that number. Here's the thing where most people spin out. There should be a break mechanism on the things that you're doing […] a culture of experimentation, and objective critique around 'Is this the right thing to be doing?” Welcome back to our two parts episode with Rachel Lee Nabors, we continue with talking about developer portal stakeholders and your metrics based approach.