Inside Tech Comm with Zohra Mutabanna
Inside Tech Comm explores how technology, content, and the changing workplace are reshaping technical communication—and the people behind it. Through candid conversations with practitioners and thinkers, the show looks beyond tools and trends to examine how the work is evolving, how people are navigating that change, and what it means for the future of the profession.
Inside Tech Comm with Zohra Mutabanna
S8E6 Documentation Is Becoming the Interface
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Documentation is not just support material anymore. It is quietly becoming a product surface, a competitive advantage, and in many cases the raw fuel that AI systems use to help customers get work done.
I’m joined by Nico Berry, a former technical writer who now works in product management at Contentful. We start with his move from technical writing into product management and the skills that carried over: technical storytelling, interviewing, and the ability to translate complexity across audiences.
But the conversation quickly moves into a bigger question: What happens to documentation when AI becomes another way users interact with a product?
We unpack headless CMS, structured content, APIs, and MCP in practical terms. Nico explains why separating content from presentation matters when that content may be consumed not only by websites and apps, but also by LLMs that interpret it for users or take action on their behalf.
That shift makes context, structure, and governance increasingly important. If AI becomes another interface to your product, the quality of the experience still depends on the quality of the information underneath it.
We also get into the tradeoffs: hallucinations, lower-quality AI-generated content, brand trust, data privacy, and a concern that gets less attention than job replacement—the fragmentation that comes from being expected to do more simply because AI makes us faster.
This conversation is ultimately about more than using AI to write documentation. It’s about what documentation becomes when AI starts consuming it.
Use discount code ITC26 to save $200 on registration for LavaCon 2026.
Season Welcome and Sponsor Note
ZohraHello friends and welcome to season eight of Insight Techcom with Zahra Mutabana. This season features a collection of conversations that explore the ideas, challenges, and opportunities shaping technical communication today. From AI and content strategy to leadership, product thinking, and the future of our profession. Each episode offers a fresh perspective from people doing the work. Let's get started. Season 8 is proudly sponsored by LavaCon. Use discount code ITC26 to save $200 on registration for this year's conference. I'll share more about LavaCon later in this episode.
Meeting Nico and His Career Shift
ZohraHello listeners.com with Zora Mudabano. I have with me here Nico Berry. And I was very fortunate to actually get introduced to him through Manny Sova. We were at the Right the Docs conference in Portland back in May. And Nico and I started talking, and then I kind of learned that he's moved from technical communication into product management. And that just got me curious. And I invited him to join my show. And he was very kind enough to accept that. So here we are today. And welcome to my show, Nico. How are you doing?
SPEAKER_00I'm doing well.
ZohraNico, tell us a little bit about yourself, your journey, and what brought you where you are today.
SPEAKER_00Hi, I'm Nico. I am not actually a technical writer anymore. I just given like the quick background on this, I got into technical writing kind of on accident after failing to be a journalist. And from there, I found that contrary to you know the famous blog I'd rather be writing, it was actually every other part of the technical job that I enjoyed most. And so what it turned out was that there was a lot of overlap between that skill set and product management. And so eventually somebody at the company I was at had the bright idea to give me a shot as a product manager, and it just kind of went from there. But I remain really interested in technical writing. I've remained in this developer experience niche. And really, I'm gonna say 60% of developer experience is just having good documentation.
ZohraThat's such a refreshing thing to hear from product management. So uh thanks for that, Nico. What were the skills? You said that there were skills that you kind of leaned into. For somebody who doesn't know what product management is, what would you say were the transferable skills and a little bit about product management itself in your context?
What Product Managers Do Daily
SPEAKER_00So, what is a product manager? The way that I would describe the core responsibility of a product manager is the person who tells developers and really the business in general what we need to build next and why we need to build it. And so within that, there are a number of things that are complemented quite well by the technical writer's skill set. The first one is just clear technical storytelling, being able to articulate to three, four, five, however many different audiences within a company, what you are about to build at different levels, whether that's talking about the low-level stack to developers or just talking about the highest level business implications of it to an executive leadership team. And beyond that, the other core part of the job is talking to people and getting information from them that may not be easy to initially understand. And that's another very common skill set for technical writers.
ZohraThat's right. So as you kind of moved into product management, how long have you been doing this? Because I kind of want to understand the pre-AI and the post-AI shifts that may have happened with your role in product management.
SPEAKER_00Yeah, definitely. So I've
AI Tools That Speed Up PM Work
SPEAKER_00been in product management for about five years. I got in right as the pandemic started, and then have been in here to some extent for a half a decade since. And the pre-AI and post-AI like phases of it have been, I'm trying to decide whether they have been all that different, honestly. I think that like the core role of the product manager has not really changed all that much, but AI has provided a number of tools that complement it, accelerate it, and provide ways to do it with a new skill set and a new tool set. But really, at the end of the day, I've seen a lot of people discuss, you know, product managers taking better to AI than many careers do. And I think that a portion of it is just that like a lot of product management is already talking to somebody else, having them make something on your behalf, and then having to stand by the results of it however it comes out. And I think that that's some a reason why product managers are a lot more comfortable with like what AI does than a lot of other career paths, where AI produces slop in a lot of cases. It takes a lot of work to get it to do what I would describe as C work for a developer or a technical writer, but I've already got practice standing by work that I can't really like do myself.
ZohraAll right. You mentioned accelerated and with the new tool sets, and that's something that I'm also experiencing as a technical writer at the moment.
Hallucinations Verbosity and Doc Quality
ZohraThat said, with the acceleration, have you noticed a drop in quality of content? Has it gotten better? What has the outcome been with AI? I'm assuming I think I'm kind of jumping ahead. So is there AI? Are the technical writers on your team using AI tools to generate content? And if so, what has the outcome been?
SPEAKER_00Yes, uh, my team is using AI tools not for generation of content necessarily, but for a lot of parts of the research and initial drafting phase. And the reason we don't just go straight into directly generating content so far has been because the quality is just lower. Like, hallucinations are real. There isn't a like, unless you are testing the uh claims that your AI assistant is making, you have no way of knowing if it's correct or not. It's also just really verbose. Like, it's a pain to read. Also, another part of it, too, which is just that like in a lot of cases, the problem that like certain sets of docs have is just that they have gaps in it, that there are more surfaces of the product to cover than you have technical writers to do it or time to do it or the ability to prioritize. And I think that there is a determination that like product managers and technical writers need to make about certain portions of the docs of is something better than nothing.
ZohraYeah. So in your current role, with the acceleration that's happening in most tech companies, what I'm hearing and what I'm experiencing is the amount of work that comes your way in the pipeline has just kind of accelerated as well, exponentially gone up. And so, from a product management perspective, are you expecting anything differently from technical communicators? And is that something that your team also experiencing, perhaps?
SPEAKER_00You know, that's a question that I should have been asking myself sooner, because I don't think my expectation around technical writers changed, but I think the context around it has changed enough that secretly it means my expectation has actually changed. Like what I mean is if my expectation is, hey guys, you make sure that all of the new features are documented, and if anything is like breaking our SLAs or there's anything critical in the current docs, go back, update it, try to like make the docs better where you can, practice Boy Scout documentation where you leave things cleaner and better than you found them. But if like my engineering teams are producing, you know, 1.5 times as many features, then really the expectation has gone up too. Sort of a blind spot on my part, I guess.
ZohraTooling
Headless CMS Explained in Plain Terms
Zohrabecomes important as well. And that's how I think in our conversation we came up with the headless CMS. So kind of taking a step back and giving that context to our listeners, but then also talking about what a headless CMS is. Would you like to check on that question? You know, I'd like to pull apart that thread.
SPEAKER_00All right. So content management systems. I'm just gonna go out on a limb and assume that an audience of technical writers already knows what a content management system is.
ZohraThey should know that. Yes. And if they don't, then we can always use amazing tools out there to learn more about it.
SPEAKER_00So in the past, content management systems would entangle the content that you wrote, but then also tie in a bunch of technology about its presentation. The main form that this took was you had WYSIWYG editors, where as you make changes to it, it's creating this completely unreasonable, totally inaccessible HTML that is tightly bound to like the website that you are using right now. And so at a certain point in this journey, somebody had the bright idea of what if we just completely remove content from this front or from the front end and have it be stored in a form that makes no opinions about how it is presented or what technology should be presenting it. A content management system without the head on top of it, effectively. And this idea took off a decent bit. Currently, I'm actually product or a product manager at Contentful, which I believe was the first commercial headless CMS. And the reason why is because when you have this kind of agnostic approach to delivering your content, it is able to be quickly adapted to all sorts of front ends, whether that's a new web framework or your mobile app, or maybe like digital signage in a store. And now the most recent thing that has come out as like a channel for distributing content and the consumption of content is AI.
ZohraCan you give a more real-world example? I think for my benefit as well.
SPEAKER_00Tell me a little bit more what you mean, like what he wants the real world exactly.
ZohraI think you kind of give gave a good definition and you talked about how you pulled out content from uh from that framework. And then you talked about how you would publish that content, essentially. So when we are removing content, like what exactly is happening there? Like I think I'm trying to get a little more technical there.
SPEAKER_00So what is happening when you're removing content from like the front end is you are starting to move from HTML that is something like a div with a style class applied to it, or a bunch of like indicators of how something should be styled, to instead an object, probably JSON, on your server that is like call to action, and then here is the text of it, where you start to associate the content with its semantic purpose, and then how that is articulated to the customer on their client is a matter for the UI or for whatever is consuming content from the headless CMS. I hope that is making sense.
ZohraSo, you know, when we've removed content here, in in simple words, the formatting is now no longer something that is part of the content itself. That is separated.
SPEAKER_00Yeah, right? In a perfect world, yes.
ZohraIn a perfect world, right. And we are talking about the headless CMS here now, right? Where this framework applies.
MCP and AI as New UI
ZohraWith AI, with agentic workflows that we are starting to kind of interact with, I think this lends well. What is your take on that?
SPEAKER_00Yes, I think that it lends well as well. I think that being able to have content with a clear purpose attached to it rather than a purpose that needs to be inferred from the LLM looking at the raw markdown of it, it tends to be able to allow the LLM to make better assumptions about the content and articulate it in better ways to whoever is having the LLM work as a middleman between them and like the content that it's consuming.
ZohraYeah, that makes sense. So we've talked a little bit about the headless CMS and why you know people may be paying attention to it. And then we also talked about how it lends better to the new workflow.
SPEAKER_00But I think that there's like one other part of it, which is beyond content. When you start to think about software as headless, then AI can become an entirely new layer of user interface on top of it. At present, one of the beautiful things about Contentful is that we provide all of the APIs that our UI uses to customers to like write scripts for if they wanted to. And what this has meant is that if this is the approach that you want to take for using the software, you can either install our command line interface and give that to the Claud code, or download our MCP server, a model context protocol server. I'm assuming a fairly technical audience who already knows what that is. Give that to either Claude or ChatGPT or any LLM that supports it, and you can just start asking it to do things within the system, and it's able to do it.
Structured Content Context and Reuse
ZohraWhen organizations start managing content, it seems like we are starting to look at content as a reusable component instead of a document. Would that be accurate in this scenario?
SPEAKER_00Yes, and I think that part of it is a philosophic question that has emerged that every organization is going to need to deal with if they want to use their existing content and AI effectively, is to what extent is content context?
ZohraContent okay.
SPEAKER_00By which I mean to what extent can you just take your content and throw it into an LLM and expect it to produce useful results when you hand it to a user? To what extent do you need to add additional things that are not already in it? Whether that's additional meta-elements, skills, or extra instructions about how the LLM needs to do things differently from how a user would if they were doing it through the UI.
ZohraBut you consider those as the reusable part of a content chunk? Okay. And that kind of also, again, when you look at it in a more broader sense, that's part of structured content, right? Where you have content that is enriched with context.
SPEAKER_00Yes. I would say like, so just going into my personal philosophy on it, I think that content and context are almost one-to-one, but there is more to context than just content, because in many cases, context can be things that are shifting and need to be assessed in a sort of like at the moment manner. It's things like a call to an API that gets you information about the status of a system right now. Or it could be like just a running log from something else. And by comparison, content is it is reusable, but it's also something that is sort of expected to stay mostly the same unless a human comes in and updates it.
ZohraYou had also mentioned content as input. What did you mean by that?
SPEAKER_00I think I'm starting to remember. It was the question of if your documentation is effectively being consumed by middlemen who re-articulate it to users. And there is actually value in that because a user can bring additional context to an LLM for your docs. They can say things like, look, I'm not technical. I'm trying to figure out like we got two developers on my marketing team, and I want to see about updating the way that like our content management system handles publishing blog posts so that it can do a few additional things automatically. I don't speak developer. Can you explain this to me like I'm five? Or explain it to me in terms that are easier to understand. Or on the other hand, you can show up and say, I'm a developer, I'm working in the C sharp stack, don't bring me any JavaScript examples. I want this all interpreted into like the tech stack that I'm working in.
ZohraYou know,
AI Fear Fragmentation and Critical Thinking
Zohrawith one of the things that I want to ask you, and you may not have the answer, all of us are starting to feel this fear about AI. Has that been the case in your field as well? Or you feel like it has empowered you? AI has empowered.
SPEAKER_00I think that it's a very complicated question. To a certain extent, I definitely feel fear because this is a technology where when did Chat GPT get released? Was it like 2022?
ZohraYeah, end of 2022.
SPEAKER_00And it came out and was just like, oh, this can write sometimes correct code if you prompt it to. And it can like talk with you, and sometimes it gets things right, but really it hallucinates all the time. And then you cut forward four years, and you get have these like incredibly complicated or not complicated, competent models, Claude Fable, where I can just tell it to recreate some video games in the HTML5 canvas element from scratch, and it's able to get really close on a single attempt. It's crazy how far it came in such a short amount of time. And I think that there's just the sort of feeling of it's not gonna replace me now, and if the gains are modest and it sort of stalls out here, I think I'm definitely safe. But does anybody know what's gonna happen with it? And I think that that's like one source of anxiety. The other bit of fear that I have about AI is less about just losing my job and more about getting so or AI making me so fast at it that I just have my attention fragmented across 20 different projects all the time.
ZohraAnd I think that's a very valid concern because we were talking about it at work, and anybody that I talk to who's from the tech industry and using AI is fragmentation and the challenge to critical thinking. Because if you're going to be that fragmented, how does that impact your critical thinking? We've been we've been making the case that AI will free you up so that you have you can do more high-level thinking. But on some days, I feel like that's not true at all.
SPEAKER_00Definitely. And a question that I've always had about AI since like we started applying it to workflows is is there more you can do with it than make this devil's bargain of go faster but with much lower quality? And like with a question mark around, are you even going faster if the quality starts to get so low that you need to come in and correct it over and over and over again? I've got some answers there that point towards a potentially optimistic direction. And one of them is that AI does provide some novel mechanisms for being more deliberate about your thinking, but it does make you slower. I'm gonna point you I'm gonna point you and your audience towards this uh skill for Claude called Grill Me. It's worth looking up right now. It is really cool. Yeah. Grill me. Cook out, grill in, grill me. Let me grab it and I'll send you a link. Grill Me is interesting in that I'm just going to explicitly state if you use this, this will make you slower at a task. But what it does is it prompts your assistant to look at all of the context it can get about a project that you are currently on, and then just starts asking you hard questions about it, forcing you to really think through it, examine it from different angles, explain the parts where you used an abstraction, or put forward a like term that doesn't seem to be fully defined, and just make you be explicit about everything, get your thoughts down onto paper, and really force you to think through what you are proposing from end to end. And in product management, this is just absolutely fantastic because you have to deal with so many different potential stakeholders on your product, and so many like different people who want to know and make sure that their opinions are considered for the thing that you are about to build. And I've found that this is the best way for me to like start to consider those opinions up front, even when it's the end of the day and I'm burnt out and under pressure and all of these other things that starts to really narrow your field of view while you're working on something.
ZohraYou mentioned something about being deliberate that can come with AI. And that got me thinking that maybe that's true. Uh, you do have to be very deliberate when you are using AI. And in my context, right now, with the costs escalating, companies are starting to clamp down on your AI tools usage, right? You have you so you have to be deliberate right there because of the cost. So now I think about whether. Am I going to use AI? What are the scenarios? Does this consume a lot of tokens?
SPEAKER_00It depends on what you're doing with it. Like typically I know my token cost per month for like my actual day job. And there has never been a month as a product manager where I've gone above $250 of tokens spent. This is like paying through the API. And part of that is just that I don't use OPIS. I basically have reserved Opus for a few very high value tasks in touching the code base. And apart from that, it's Sonnet all the way with a few subagents that use Haiku for summary or other things that can be token intensive and not terribly difficult. I think that using this with Opus could create problems and might get you on like your accountants, you know, shitlist.
ZohraYeah.
GrillMe for Better Product Decisions
ZohraAnd I do the same thing. So the Grill Me, how did you discover this?
SPEAKER_00Honestly, every morning when I wake up, I pour myself a cup of coffee and I open up a Y Combinator's hacker news and just read through the top 20 articles for the day. And this was in there.
ZohraOkay. And can you share when you used like an aha moment when you used GrillMe and it really helped you?
SPEAKER_00Yes. So one of the things that I am in charge of as product manager at Contentful is our Model Context Protocol Server. And I started using Grill or I used GrillMe when I was putting together the state of the product document for the current quarter that we are in. And it made me talk about what we are currently seeing. It forced me to go into our data and answer some specific questions for it that ended up going into this state of the product document. And then ultimately at the end of it, the aha moment that it led me to was a question about how granular we were or weren't being with the audience of users for the model context protocol server. Because it main goal that I have had in building it out is this is a strange way to put it. When you work in like B2B enterprise SaaS, you're making some of the most boring software in the world. And at best, the like good human moments that make it worthwhile are when you can make something that makes somebody's job so much easier, or like gets rid of some task that annoys them a lot, and they go, oh, so much better. Providing the model context protocol server and Claude to like the common users who aren't developers that use Contentful, I could visibly see their faces light up and like joy hit them, that they had hit a point where they could start to describe what they wanted to have done with their content strategy and their marketing strategy, what they needed to do to maintain it mechanically through our software. And it can just happen without them needing to remember how the UI works, how to like click through 20 different nested menus and grapple with the not great parts of having a UI that can serve highly complicated use cases. And so what this early realization led me to was going, how can we get this into the hands of more people who are not developers? And ultimately, what my grill session led me to was, and I hope that this isn't just like super obvious, that the actual thing that needs to happen is making all of these different governance cases to control what an agent is able to do and be able to hand it out to people with like the confidence that even if they don't know what they're doing, they can't cause serious damage to like the setup for your company. And really focusing on providing that to developers with clear messaging about why they should care about it and what they should be building for their internal team.
ZohraAnd yes, it's known as the Fun Conference 2, with networking events, therapy docs, comfort llamas, storytelling, and karaoke. Use discount code IDC26 to save $200 on registration. Yeah, that's such a valid point. And
When AI Content Hurts Brand Trust
Zohraone of my questions was how are customers thinking about AI generated content, or if they are being, you know, through your tools, being uh using the MCP protocol and able to generate their own content using your content as input, what has their experience been like? So this example that you shared kind of illustrates that aha moment for them as well, probably.
SPEAKER_00And I want to be clear, the thing about this is also something that applies to technical writing as well, but a lot of the customers that we have are content writers and marketers. And when they ask me about using AI to generate content, I actually encourage them not to do that. Because AI generated content, unless you are very careful with it and very deliberate about having lots and lots of highly refined content that you're inputting to it, and like you're basically just generating little variants for different markets, different times of year, just like the last 10% of doing content strategy. If you don't do that, the content will come out sounding like AI. And there's a decent amount of evidence showing up in the industry that content that sounds like AI just makes your brand less trustworthy.
ZohraAnd that's where the human in the loop, human oversight, governance, transparency, all these things start to matter. Have you had any pushback from customers at all that they don't want any that is?
SPEAKER_00Yes, definitely. There are plenty of customers, like I would say a decent third-ish of our customer base just doesn't want to touch the stuff. And you know, we accommodate that and provide a thorough off-switch for it. I think that both approaches are valid depending on what you are trying to accomplish and which market segment you are in. Like there are places where using AI makes a lot of sense because what's going on is you're writing content that you need to get to rank on long tail SEO keywords. And a lot of what you need to do is just take some core themes of your content and have dozens of variations. It's things like if you are a food company and you make hot dogs, there are only so many ways you can provide recipes and instructions for making hot dogs, but you can always find ways to connect it into like which season is it right now? What holiday is coming up next? Things like that.
ZohraOkay. So the resistance that some customers thought of the customers, so I understand that market segments definitely will determine if they want to use AI or not. But are there customers who you think are in a segment, market segment that would benefit, but are pushing back perhaps because either they are not AI fluent or just AI resistant and being more old school?
SPEAKER_00I
Data Privacy and Regulated Rollouts
SPEAKER_00have definitely seen customers who are like that. Part of it is just I think that there is some deliberate efforts on the part of frontier model providers and the providers of like LLM tools to obfuscate what exactly is happening under the hood. And this can start to like make it intimidating, complicated, and feel a lot less trustworthy than it actually is. And like I think that that results in some people just not really knowing things like what is happening with my data as it is being passed around through this stuff. Right. Where in some cases, like, depending on which provider you're using, you may either genuinely not know, or it could be things like, oh, this what's happening is the data is being routed through AWS, it is going to hardware that is owned by Amazon and governed by them and adheres to their privacy policy, and it gets run through an open weight model that is hosted on their hardware and returned back to you.
ZohraI see. So those kind of scenarios can turn off the customer. And and I don't think that's fair, especially when it comes to protecting their data.
SPEAKER_00Yes. The other thing is just depending on like what software you're using, there are things like there are situations out there where the stakes for getting something wrong are very high, and AI makes it easier to get things wrong.
ZohraYeah. So I was going to ask Philip the question that I had for you. Are there customers who are very gung-ho about using AI, but where you feel as a product manager that probably their adoption should be slower and you might want them to go slower?
SPEAKER_00I'm going to say that has never been something that I have personally encountered, thank God, because some of the customers I work with are in the healthcare niche. Some of them work in government, some of them work in finances, and there are a lot of places where the rollout of AI is very slow, very deliberate, very conservative, and they are making like every or taking every pain possible to get it right. And I'm so happy that that's the case because there are actually substantial downsides to them getting it wrong.
ZohraI rarely end up thinking about these different industries, but I'm thinking about education where things definitely would be probably slower too.
Avoiding AI Without Clear Goals
SPEAKER_00The thing that I think is a lot more common that I see is companies that are going headlong into AI, but don't really have a specific goal apart from use AI. And like I'm still not sure what to think about that because this is an expensive technology to roll out. I'm sure we've all seen some headlines about companies racking up multimillion dollar clawed bills. To some extent, it's just unless you know what you're getting from this or have a very good use case for it, it feels wasteful and disruptive to roll that out to a bunch of different teams and also then pay for what is effectively getting you nothing. At the same time, I've also seen some product managers make the argument that you actually kind of need to have this phase of just getting people to start using it so that they can discover the thing that it is valuable for in their workflow.
ZohraYeah. Probably I would say three years ago, when AI was new, generative AI was new, it made sense. But I think if companies are a little late in the game, they probably need to have some controls, some guardrails, and not just free for all. I think that train has left the station. But I would be concerned equally about that.
SPEAKER_00Yeah. I kind of agree, but also, man, in a highly regulated industry, three years is no time at all.
ZohraRight. In a highly regulated industry, yes. But in a non-regulated industry In that case, you're right. Yeah. Because in a regulated industry, anyways, things are going to go slower. And there are guardrails, but it's the non-regulated industry where I feel that education needs to happen, that training needs to happen, and guardrails absolutely need to exist. Because again, if you have customer data and if you're not considering that aspect, that would be a challenge, I think. We have covered good ground here. This is just a question that I'm curious about. And let me know what your take on that is. What
Letting AI Write User Stories
Zohracontent practice or even product management? I mean, you've been in product management, you said about five years. What was a best practice five years ago that's no longer a best practice, in your opinion? As a product manager.
SPEAKER_00Writing your own user stories for sure is one of them.
ZohraOkay. Yeah. How do you feel about that?
SPEAKER_00I'm so happy that that's gone. Like, that's one of those ones where it's like every once in a while I'll see a developer like complain about Jira on the internet and say something like only product managers like Jira. And I'm like, no, bro. Saying product managers like Jira is like saying that people with leprosy like leprosy. It's not something we use because we like, it's a disease, but I see.
ZohraSo you'll feel like liberated that Yes.
SPEAKER_00Having something that can sit between me and that system and offer a layer of abstraction that helps like remove this mechanical task and can just take the higher level goal and enumerate out all it into all of these different chunks that mostly have like boilerplate between like what is in the initial requirement document, the initial spec document, and what needs to go into stories. That's been a dream come true. And it is a place where like a couple of hours of my day on average have been freed up, and I've been able to put it right back into stuff that feels better to do because it's more impactful.
ZohraYeah, that's a great example.
SPEAKER_00The
Claude Code and Docs as Code
SPEAKER_00technical writer version of this that like we've started to play around with is using plug code as a way to deal with just the nitty-gritty of the Git-based workflow for Dox's code. And that's an interesting spot because like in technical writing, there's a number of technical writers out there. I was one of them at the time, or when I was a technical writer, who love Dox's code, who love the fact that you get to get down into like the details, go into the command line, learn Git, play around with different markup languages, define your own components, run the static site generator locally, make a GitHub action for it, and like do all of this really technical stuff. And then there's a lot of technical writers where that's just like the miserable part of the job. It's something they're not interested in that stresses them out, that they view as like, uh, okay, if I mess this up, I'm gonna have to go contact like a developer Googling a solution for 30 minutes. And for those people using Claud code and just saying, hey, I got these updates that I need to make, let's do it.
ZohraOh, yeah, definitely. The initially I was a little hesitant to use that, but I'm one of those who likes to be doing Dockest code and be in the developer workflow. So I feel very empowered. But yeah, you're right, that it it can empower somebody who doesn't want to do this as well. It kind of deburdens them.
SPEAKER_00I think that this is one of the parrot, like I don't know if it's actually a paradox. I'm gonna call it a paradox. One of the conundrums of AI right now is that for people who are very, very advanced and established and good at a trade in like tech, the benefits that AI provides can be like a lot lesser because search, looking things up, well, if you've already got your programming language and the systems you're working in and your code base committed to memory because you've been doing it for years, it starts to be less useful. Then there are people who don't have any of those things who can benefit a lot from having like a little machine assistant who's able to spin up like a prototype or handle the Git workflow for doing Dox's code stuff. And it frees them from having to commit a lot of time to learn these skills that are kind of ancillary to the thing that they actually do for their job. But also because they don't learn those skills, they're they have less in place to defend them when things go wrong. I'm still figuring out what to do with all that.
ZohraThis
Final Takeaways and Listener Requests
Zohrahas been, for me, Nico, a good conversation. You've really picked up on a lot of good stuff. And I came scripted, but I think we went off the script entirely.
SPEAKER_00This happens when I start talking about AI with people.
ZohraIt's which is great. That's my goal. The script was for me, not really for you, because I knew this would end up being a great conversation. I just had to prepare myself to be able to understand. And thankfully, as you were giving these examples, since I've been in these tools, I was hopefully able to ask some intelligent questions of you.
SPEAKER_00Yes, definitely. And I'm going to And hopefully I gave intelligent answers.
ZohraOh, that was, I mean, that was not even up for questionnaire. I that was anticipated. But thank you, Nico, for being on the show and sharing your experience and your insights with us because it's definitely helped me. And you've challenged me to think about AI differently as well on a personal level. So thank you again. Have a lovely evening, and uh hope to run into you at another conference.
SPEAKER_00Absolutely.
ZohraThank you for listening to Insight Techcom. If there's a guest you would love to hear on the show, please let me know. And don't forget to follow and catch every episode on your favorite podcast app. See you next time.