How AI Is Reshaping Documentation Workflows – Ankur Parekh, SnapLogic

Transcript:

• • • 01:17 – 02:52 Ankur’s Journey into Technical Writing

      Gowri Ramkumar: Good day, everyone. Welcome to Knowledge Base Ninjas podcast.

      With me today, we have Ankur Parekh, Senior Technical Documentation Manager at SnapLogic.

      Hi, Ankur. How are you doing today?

      Ankur Parekh: Good, Gowri. How are you?

      Gowri Ramkumar: I’m good, I’m good.

      So great to see you having your coffee for the day. And I’m sure it’s going to be a brilliant chat in the next 15 to 20 minutes.

      20 years in this field! Talk me through how did you get into this? And what inspired you to choose this as your career path? And how are you enjoying?

      Ankur Parekh: Sure. So I started about maybe 25 years ago, almost, as a college, you know, fresh college grad got into copywriting. Actually, I was in advertising as a copywriter. And shortly thereafter, you know, the dot-com boom happened. So, you know, obviously, I wasn’t the one who would want to be left behind, so to speak. And, so I got into web content. And from there on, in 2004 2003, roughly, went into mainstream, technical writing with the startup. So, yeah.

      So, from copywriting to web content to documentation, that’s been my journey, and ever since, have been kind of, you know, continuing with documentation, and happy to be here.

      Gowri Ramkumar: Right. So 25 years. So much. So much to talk. Let me try to get a few bits from you in the next minutes.

    • 02:52 – 05:10 How AI Is Reshaping Documentation and Blurring Roles

      Gowri Ramkumar: Now, in some, lots of stuff changed from when you started, how you started to where you are right now, right?

      Now, in your view, what’s the biggest way AI is changing how documentation is created and then how it’s maintained? So, I’m going to start with AI to you first.

      Ankur Parekh: No. Right? I mean, that’s the hot button topic, actually. Right?

      It’s very exciting. At the same time, I think, we are all under pressure to leverage AI as much as possible. And I think a lot, a lot has been already said.

      What I would say is, looking at the future, I expect AI to obviously change the way we are documenting today, right? Essentially, it will be a powerful tool for us to optimize our process, improve productivity, and the quality of the deliverables. Right? So that’s what I feel like, and importantly, I also think that it is going to blur the various roles, adjacent roles, that we find ourselves into.

      So, for instance, you know, you will have an instructional designer, right? Doing training videos. You have a UX writer doing UI content and the technical writer doing user manuals and admin guides, and all. So all these rules are going to blur, in my opinion, because AI will help you.

      For example, if I have one valid input in terms of documentation, right? In whichever form it is, I can use that single piece to not just create multiple, you know, channels in terms of documentation that we did previously, like HTML or PDF, and stuff like that. Rather, you can create various types of deliverables, like you will be able to create videos, you will able to do training videos, how-to guides, tutorials. So a single piece of content would be able to spin various forms of documentation. You know, that’s what I feel we are going towards. And we have already kind of seen early signs of that. So these roles, then the traditional roles, are going to get blurred.

      Gowri Ramkumar: Right. Yeah.

    • 05:10 – 07:25 Tackling AI Accuracy Challenges

      Gowri Ramkumar:It sounds all good and nice. But then there are also, fear that AI tools can occasionally hallucinate the contents, right?

      So, how do we know, or how can we ensure that the accuracy and credibility in AI-generated documentations are maintained?

      Ankur Parekh: Right. So one thing is for sure, right? The human also always has to be in the loop. So that’s not going to change really, Gowri, for us. Be and you know it’s like if you want good coffee, you have to put good coffee beans. Otherwise, you know, it’s garbage in, garbage out. So the same goes for AI. The input, the quality of the inputs that we provide the AI tool with, has to be top-notch, right? In terms of accuracy, completion, etc. If you have that, if that’s validated, then you can expect less hallucination. And obviously, over time, as the models get more mature and trained, I think the level of hallucination will drop.

      And I’ll give you an example here.

      Gowri Ramkumar: Sure.

      Ankur Parekh: We are migrating today from, say, you know, HTML to data, right? We have been doing that since last year, and we have been leveraging LLMs for that, right? And what we are finding is that sometimes, for example, while migrating the content, it changes the tags, right? So it applies the wrong tags or it drops some form of the content, you know.

      Gowri Ramkumar: Okay.

      Ankur Parekh: Because if especially when it comes to complex long-form documentation like we have at SnapLogic, it does that. And so what we have kind of normalized to is, hey, let’s kind of use the LLMs to chiefly migrate uncomplex, shorter forms of docs, right? To begin with, and then we are trying to kind of push the boundaries and trying to see if we can migrate even more complex topics, especially tables.

      Gowri Ramkumar: Yeah.

      Ankur Parekh: It kind of, you know, misfires many times.

      Gowri Ramkumar: Right. Okay.

    • 09:11 – 13:13 The Promise and Pitfalls of AI in Documentation

      Gowri Ramkumar: Now, you spoke about your current project, where you are moving from one system to another, and then you are using some parts of the AI in that as well, right?

      Now, I’m going to ask you what excites you the most about the potential of AI, and also what concerns you the most?

      Ankur Parekh: So the potential is, you know, if we track the LLM’s journey till today, in the last couple of years, it’s come a long way. And I remember, for instance, you know, I used to be able to, one year ago, ask it questions around, you know, some of the security-related tests that we have to give, training that we have to take internally. And it used to not be able to provide with answers, right, when I used to give them the options. Today, when I do that, it is able to precisely tell me what the correct answer is. Right. And, so that just shows us, you know, how quickly this technology is evolving. So, I’m excited with that for sure.

      Obviously, the flip side is that we are going to be expected to do, to deliver much more, and I’m sure demonstrate how we are leveraging AI both internally for in terms of process, as well as externally the customer, right.

      So, for example, we are, you know, for our own documentation portal, we are leveraging AI to build an AI assistant for the documentation portal. Wherein, you know, we now introduce an AI-powered search, summarization feature. And also, obviously, conversational text, right? So that the users don’t have to necessarily dig through everything. They can just have like a conversation with the chatbot.

      Gowri Ramkumar: Correct.

      Ankur Parekh: And one last thing that I’m trying to suss out here is if we can have changelog, right? So what, because that’s one of the requirements or one of the ask that customers have asked us that, hey, can I know what’s changed in this particular documentation? Right.

      Gowri Ramkumar: Yeah.

      Ankur Parekh: Over the last, maybe six months. And so that’s something we are trying to figure out because we use GitHub as a repo. And I’m like, okay, GitHub knows what’s changed. Can we, you know, cohesive, seamlessly seamless way, communicate the same to the customers?

      Gowri Ramkumar: Right.

      Ankur Parekh: Right. So that’s, so these are some of the use cases. For example, we are experimenting with externally, right?

      And internally apart from you know, we are already leveraging AI for some of the task within the documentation lifecycles. So starting with research, PR review and editing.

      Gowri Ramkumar: Right.

      Ankur Parekh: So, these are the tasks that we are already using AI for. And we are obviously improving over time, right. And from a future state perspective, obviously, you know, we are trying to see if we can leverage AI to generate original technical content.

      Gowri Ramkumar: Okay.

      Ankur Parekh: Right. Providing it all the inputs that we have, assets like JIRA ticket, the business use case, test plans, code, right? If I give all of that, which traditionally we used to kind of go over everything, all this to understand the feature and the functionality and the benefit that accrues to the user. So, can we do that, if I gave the same input to the AI to the LLMs, would it be able to provide me with baseline draft content? Which is fairly technically accurate, such that I can quickly validate it, thus boosting my productivity.

      Gowri Ramkumar: Right.

      Ankur Parekh: That’s where we are going towards.

      Gowri Ramkumar: Okay. Got it. Got it. Wow, that’s a lot of experiments and a lot of exciting things happening for your Ankur. Good luck with that.

    • 13:13 – 19:11 ⚡Rapid Fire Round

      Gowri Ramkumar: Now, let’s move on to the rapid-fire round questions.

      So I’m sure you must be reading a lot of resources, especially as you are trying multiple things. Can you point out a few resources, particularly on the documentation space?

      Ankur Parekh: Few resources in relation to AI?

      Gowri Ramkumar: Maybe some blogs, or any books, or any sites that you came across that really made you think, wow.

      Ankur Parekh: Well, that’s a good question.

      And I don’t think I’m going to be able to answer in a rapid way. What I would say is, you know, I obviously just knowing how quickly the technology has evolved, right? If you think about it, the last, you know, if I think about my own career, last 20 from going back, you know, 20 years, the first ten years, the rapid, the change, and the rate of change versus the last ten years and the foreseeable future, the rate of change is going to be exponential, right? So what that means is we have to obviously keep up with, you know, we have to engage with the community. And when I say community, I’m, you know, talking about wherever we can get knowledge, right? So I attend conferences, webinars.

      And so for example, earlier this year I attended the Convex conference in San Jose. That was obviously very useful. I also attend, there’s a roundtable that happens, that is organized by Oberon Technologies there somewhere, I think in the Midwest. So they do these roundtables all over the country for technical documentation leaders.

      So I think that, you know, once in six months, then that is the another, you know, I think, CIDM, Right? CIDM is basically, Comtech services. Right. So they are also, I think somewhere in the Midwest. So they, I’m a member of that and they organize also webinars. So that’s another, on hot topics, basically, right on topical things. And so I attend that as well on a regular basis.

      So these are some of the things that I do to kind of keep up with what my peers are doing and where the industry is headed. I think, you know, we have to keep experimenting, which we always did as content professionals, but I think we need to do that exponentially much more if we want to be, you know, if you want to stay relevant.

      Gowri Ramkumar: Great. So basically you attend a lot of conferences, try to do a lot of networking during the conferences and events, and stay on top. Nice.

      Now, what is that one word that comes to your mind when I say “Documentation”?

      Ankur Parekh: I think education, right. User education.

      Gowri Ramkumar: Okay. Okay.

      A piece of advice you would give to your 20-year-old self?

      Ankur Parekh: Well. Never assume.

      Gowri Ramkumar: Right. Okay. That was very rapid fire. Yeah, yeah.

      Ankur Parekh: Kind of.

      Gowri Ramkumar: Because you started your career straight from university, straight after graduation, right? So how did you, how did you take it in your early days?

      Ankur Parekh: Well, when you say early days of my documentation career?

      Gowri Ramkumar: Yeah.

      Ankur Parekh: Yeah. Okay. Well, early days, you know, it was trial by fire, right? I essentially that time, there was not many courses around technical documentation and all. So I relied on the, there was this international mailing list, at that time. And so I relied on that. And, you know, there were my counterparts in the US at that time because I was in India, they were teaching me. You know, I think it was RoboHelp. Okay. Suppose, over the IM, right? And that’s how I was picking up. So getting my feet wet in technical documentation was not easy. But I think it was fairly rewarding, from that perspective. So…

      Gowri Ramkumar: Right.

      Ankur Parekh: Yeah. And there were people, at that time telling me that, hey, why don’t you just, you know, like technical writing institutes, right? That why don’t you just join us? You know, we will pay you a stipend, which basically means that they put you on a client. And then they pay you from that. And I said, no, I don’t want to do that. I already have the writing skills. I just need to know the tools and understand the nuance of the medium, right. And, yeah, it took me like a couple of years to completely immerse myself, but after that, it was just fine.

      Gowri Ramkumar: Great. Yeah. I’m assuming that’s what you would have done. Never assumed anything, but went deeper into and took your time to analyze and understand things.

      Ankur Parekh: Right. Yeah. Yeah. You can never assume in every sphere of life, right? Go out, go there with an open mind. You know, whatever it is, your life, personal life, or work life. So that’s what I would say to my 20-year-old.

      Gowri Ramkumar: Alright. Super. So Ankur, I know there’s lots going on and, you’ve taken this time to share your experiences with the audience, and it really means a lot to us. It was great connecting with you and grasping some of your minds around the space.

      All I can say is that wish you all the very best. And, yeah, well, we hopefully look forward to hearing from you in a couple of years on the successes of what you’re currently doing.

      Ankur Parekh: Sure, Gowri. Thank you. And it was, great talking to you. It’s an absolute pleasure to meet you as well.

      Gowri Ramkumar: Thank you! Take care.