Podcast: Play in new window | Download
Subscribe: Apple Podcasts | Spotify | Android | TuneIn | RSS
One of the most engaging aspects of generative AI products is their conversational interfaces. This has led many content designers working on AI products to develop skills in conversation design.
Sharon Ni works on both conversational AI products and script-driven chatbots in her content design role at Cisco. She has developed her conversational design and technical AI skills by attending conferences, hackathons, and other events, by networking extensively, and by experimenting constantly with AI and conversation tech.
We talked about:
- her work on chatbots and AI tools at Cisco
- an overview of the content design guidance chatbot she built
- her addition of “conversation designer” to her resume
- the evolution of the people ecosystem she works in, which now includes more engineers and data practitioners
- the professional development that she’s done to prepare her for working with AI and collaborating with her more technical collaborators
- how participating in hackathons and other events has helped her advance her AI skills
- some of the tools she uses in her work, including spreadsheets, Miro, and Voiceflow
- her personal interest in building chatbots and how it’s helped her in professional work
- the content design repository where she stores the conversational content she works with
- how she helps her colleagues understand how to best use AI
- her new responsibilities around assessing the technical feasibility of
- her advice to “just do it,” to start building your own AI projects and connecting with others who share your interest
Sharon’s bio
I love writing products. I hate writing about myself. So here’s five quick things about me and my work in AI:
- I’m a content designer at Cisco. Currently working on the Cisco AI assistant and Cisco.com chatbot.
- I like trying and building different chatbots myself – I recently built a content style guide chatbot that can help people review their copy and find guidelines.
- I’m a fierce advocate for content research and like to use data to inform my content design decisions.
- I have a background in Psycholinguistics and received a master’s degree from Middlebury College in 2023.
- Huge fan of this podcast.
Connect with Sharon online
Video
Here’s the video version of our conversation:
Podcast intro transcript
This is the Content and AI podcast, episode number 35. One of the main attractions of generative AI products is their conversational interfaces. This basic characteristic has drawn many content designers into the adjacent field of conversation design. In her work on chatbots and conversational AI products at Cisco, Sharon Ni has applied conversation design techniques and also learned a lot about the engineering side of AI, sometimes even advising her colleagues on the technical feasibility of their product ideas.
Interview transcript
Larry:
Hi everyone. Welcome to episode number 35 of the Content + AI podcast. I am really delighted today to welcome to the show, Sharon Ni. Sharon is a content designer at Cisco, is doing really interesting stuff with AI and other technologies there. Welcome, Sharon. Tell the folks a little bit more about what you’re up to these days.
Sharon:
Yeah, hi Larry. Very nice to meet you and excited to be here. And as you mentioned, I’m currently working on Cisco AI system for security, which is part of the Cisco AI ecosystem. And I’m also working on a chatbot that’s on the cisco.com website right now.
Sharon:
And other than that, I am also working with the Voiceflow team to build an AI powered content design, style guide chatbot that can help our design partners to find the right guidelines and also review copy based on the guidelines, basically. It’s not going to write the copy for them, but it will provide recommendations based on the good examples and bad examples that I fed into the chatbot and also the templates. So yeah, that’s what I do.
Larry:
Well, it sounds like you’re definitely earning your paycheck, at least three major things going on there. I would love to start with the content design guidance chatbot that you mentioned, because that’s like… I think that’ll be of interest just probably a lot of people are working on similar kinds of things. Can you talk a little bit just in general about… You mentioned that it’s not so much doing, writing for people, but it’s more like style and voice and tone and stuff. Anyhow, can you talk a little bit about how that chatbot works?
Sharon:
Yeah. So basically, I injected all of our guidelines into this chatbot. I kind of rewrite it because you can’t just put the same… the guidelines into the chatbot. It’s not going to recognize it very easily.
Sharon:
And so I work with the Voiceflow team. They help me to write the code part. And then right now I’m just adding more examples from our product, the copy and our product, and also some good examples and we also need some bad examples so that the AI will be able to recognize it and learn from it. And also the templates that you have to provide with the… what kind of response you want this chatbot to produce in a certain format. The reason why I wanted to create this was because we always get a lot of repetitive questions from…
Sharon:
During our office hours or in our help channels, people are asking about whether or not it’s okay to capitalize certain words or sentences. And also they’re asking about some words that’s already… they’re in our guidelines. So that’s why we wanted to create this chatbot so that people don’t have to look through our guidelines. They can just type in using natural language and to find the right thing that they’re looking for. Yeah.
Larry:
Right. And you’re building that with Voiceflow. And it’s interesting, you still have the job title: content designer. But you’re doing an awful lot of conversation design.
Sharon:
I know. I know.
Larry:
Working with Voiceflow and all that. Was that a new skill to you? Because you’ve been doing this, what, a year or so that you’ve been working on these chatbots?
Sharon:
Yeah, like a year.
Larry:
So you’ve kind of upskilled to become a conversation designer as well as a content designer?
Sharon:
Yeah, I think so. And I think I started calling myself conversation designer very recently, because I feel like all my projects right now are AI or chatbot related. But also, at the same time I feel like the conversation design work that I’m doing, just wanted to be clear that might be different from what other companies are doing or other content designers are doing.
Sharon:
But I think basically right now I’m doing a lot of the writing for AI and also the writing for chatbots. But also, at the same time I’m working with a lot of design team, marketing, and also sales team to just think of those strategies for AI. So it’s more like a new experience to me, but I find it really interesting and I had a lot of fun with it.
Larry:
Yeah. And you’re reminding me of… There is… It seems like generalists, or not so much generalists, but people with versatile skill sets are really going to thrive it seems like in this age of AI, because what you just described… And not just skills, but also the ability to collaborate with new and different people. Like the conventional content design roles, there’s the product and engineering and design colleagues where you just mentioned that you’re working… Well, this has to do with the nature of the products. You’re doing the sales and marketing folks.
Larry:
But you’ve also mentioned, I know in your AI work you’re working with machine learning engineers and data scientists and stuff. Can you talk a little bit about how the people ecosystem around you has changed over the past year?
Sharon:
Yeah, yeah, definitely. I would say in the past I’ve never really worked really closely with the engineers in the past. Just for our team, we mainly work and support our designers. We’re more like a service. And also, because we have a super, super small team. We only have three content designers in our team. So a lot of times we’re not the one who actually created the copy at the very beginning. We’re more like a reviewing, we’re helping them to review and also to edit their copy. And also we have our office hours and help channel to help them answer UX writing related questions.
Sharon:
And right now, I think I’m more embedded in those AI projects from the very start of the project. And I’m doing more than just writing. I know people are talking content designer, only 10 or 20% of their time are doing the writing work, but right now I feel like it’s less than that.
Sharon:
We’re actually doing more thinking than writing, which is really interesting to me. And I’m in this AI design team and we have our designers and we have our engineers, machine learning experts, and also AI experts and data scientists.
Sharon:
We work really, really closely together because what we’re doing right now is we’re all trying to figure out together. We don’t really know what we’re doing. But it’s great to be able to understand and also to learn from other people and to learn what they do and also what they know. And I learned so much from especially the engineers about AI and especially the technicality side of AI and also the limitation of AI, what we can do with AI, what we can’t do with AI. So I think this is super important.
Larry:
Yeah, that’s one thing that has come up in a lot of my conversations, is the level of technical skill that is required to do this is a little more than a lot of conventional content design roles. Was that a challenge for you or did it come naturally or how did you get up to speed to work with these more technical collaborators?
Sharon:
Well, it wasn’t easy, I would say, but I try to take courses and also I read a lot of books around the machine learning side and also how we’re using data to inform those decisions. I think I’m coming from a linguistic background, so I had some sense of it, but I don’t really understand how this works. But I think some courses… I took some online courses, and also I watch a lot of YouTube videos. Definitely helped me a lot to understand just how AI works and also different types of AI. And I also…
Sharon:
So for me, I like to go to those conferences and also hackathons a lot. So I like to talk to people. I like to talk to the founders of some AI companies. And also I like to talk to those developers who built AI and to just understand the technical limitations of AI.
And also, I also learned RAG this way and also some other latest functionalities of AI, I would say. And I also build AI products and chatbots myself. I like building things and that’s how I learned.
Sharon:
And I think I really, really recommend people to go to those hackathons just because I feel like you’ll be able to meet a lot of, not just like-minded people, but also a lot of investors as well. You’ll be able to understand from their perspective what they’re looking for in the market and also different use cases that people are coming up with.
Sharon:
And I think there’s a really funny story. So I was at this hackathon a few months ago, and it’s a law AI hackathon, and we built this chatbot thing to help people to write contract. And there was a investor who approached me after the hackathon asked me if I wanted to build a company and to scale this product, and I said, “No,” but it was a really interesting experience.
Larry:
That’s pretty awesome because that’s a common outcome at those hackathons. And for folks who haven’t been… I’ve been to a lot of hackathons over the years. And it’s a great environment. It seems like really tailor-made for up skilling and this kind of thing. Because all these new collaborators, including maybe investors, if you were in a position where you were interested in doing something, that could be a fantastic opportunity.
Sharon:
Yes, absolutely.
Larry:
Yeah, that’s really interesting that that came up. But the other thing about that is that those hackathons, they tend to be engineering heavy, but there’s a lot of people like designers and other folks there. What are those… I haven’t been to any… My last AI hackathon was almost a year ago. What kind of folks were there? Was it similar to the folks you’re working with or was it a different crowd?
Sharon:
Yeah, I think it’s really different, but also in the same way… Yeah, you’re right, it’s mainly for developers. But also there’s going to be some subject matter expertise there and also some designers, not a lot, but some designers. So a lot of time I have to be the designer. I have to work with Figma and I have to draw all those prototypes and stuff like that. So it also helped me to build my skill in that area. And also, I feel like it’s really interesting to just not doing actual work, but to talk to people and learning about their experience.
Sharon:
Because I realized that a lot of people who would participate in those hackathons, they’re doing hackathons for… I don’t know if they’re doing this for life. But I heard a lot of people would fly to different places just to participate a hackathon and just to gather ideas and so that they can probably start up their own company or something like that. But I think this is super interesting like, talking to people there and also being able to understand what they did in the past and also how they started working on AI. I think this is really interesting to me, and I also made a lot of friends there. So, yeah.
Larry:
Yeah. Well, no that’s like in, well, the people stuff. It does… I want to revisit that a little bit too. These new collaborators, especially the more technically oriented people.
One of the things we talked about before we went on the air was sort of your communication artifacts. You’ve got a lot of new collaborators and a lot of new communication that needs to happen. You just mentioned Figma, that’s probably one tool you’re using, but can you talk about… There’s so many things you need to communicate. You’re discovering stuff as you talk to the end users, but you’re also talking to your collaborators. How do you share all this information with folks?
Sharon:
Yeah. So right now we’re using a spreadsheet. I feel like that might be a better way to do it. But still, we’re going into very traditional way of documenting the content, is through a spreadsheet kind of thing.
Sharon:
So we have a spreadsheet and also we have a Miro board for the decision trees, for the flows. So we would actually show our stakeholders, both stocks together for them to understand how this conversation really works or looks like behind the scene.
Sharon:
And I would also create a prototype or a demo using Voiceflow to just show them the working chatbot would look like. And so that especially the engineers would be able to really understand what are some new flows they have to add to this chatbot and what are some things they need to update.
Sharon:
So I find it really, really helpful to have this kind of documentation and also this centralized place for all the dialogues or for all the content that I created. But also at the same time, I feel like it’s a really good way to, especially during… If we’re doing something like crits or content crits or review work, it’s really, really helpful to show your stakeholders what have you done in the past week or so. It’s also really easy for them to get feedback and to comment on this doc together. Yeah.
Larry:
Yeah. And you’re reminding me for all the other new collaborators, you still have your original content design collaborators to share with them, so good old-fashioned design crits and things like that.
Larry:
One of the other things you mentioned is that out of all of these conversations and all this artifact sharing, you begin to develop those scripts. You develop things that are sort of exemplars of how things should work. And I guess I want to quickly mention there to that you’re kind of starting both the conventional chatbot projects and your AI stuff with the same idea, right? You’re sort of learning using the old-fashioned, old-fashioned, Voiceflow kind of scripts as sort of learning material for the AI? Did I hear that correctly? Or…
Sharon:
Yeah, I think Voiceflow is really, really powerful. You can create scripted workflow, but you can also plug in the AI, the large language models part into this chatbot. And you can even, not fine tune it, but train it a little bit at least to get the results that you want and even doing some prompt engineering work. So I think working with Voiceflow really helped me to build a lot of skill sets into conversation design and also prompt engineering and even some documentation maybe. And then helped me to start learning about this, the whole AI and conversation design thing.
Sharon:
And I’m the type of people who really like to learn on the job. So I built probably five or six chatbots in the past using Voiceflow, and that’s how I got started, and that’s how I got interested in this, I would say. I started working on Voiceflow even before the Cisco AI project. And I would go to those websites and I would read a lot of newsletters, and I’ll also listen to a lot of podcasts for me to get on top of the market, I guess, and also be able to understand the logics behind those chatbots. I think some of the logics are really similar to each other. So once you be able to build one chatbot, you’ll be able to use it for some other stuff too.
Larry:
That makes sense. Now, I’m curious about those first five or six chatbots you built. Were those just personal hobbies or just-
Sharon:
Yeah. Absolutely.
Larry:
…made up… Absolutely. Yeah.
Sharon:
Yeah, those were just my hobbies, I guess. And I built this AI Resource Finder chatbot to help people to find AI resources. I think one of the challenges would be how to document all those, the links and also resources that you find elsewhere, and also to share with other people based on their level of experience on AI.
Sharon:
So a lot of times I found that people would ask me for advice on what courses to take if they wanted to start learning about AI or what articles I would recommend them to read. And I found that it’s not really effective just because I don’t really understand whether or not they have learned or they have any knowledge on AI. So I created this chatbot and they can select. So the first question they’ll see is: what’s your level of experience or expertise on AI? And then they’ll see different recommendations based on different topics.
Sharon:
There’s some articles about RAG, or some topics related to documentation or conversation design and things like that. And I think the logics behind this is really similar to the content design chatbot that I’m creating because it’s ultimately finding different information based on different level of the requirements and also based on different topics as well.
Larry:
Yeah. That’s really… Resource Finder, is that publicly available anywhere or is that something you just share with friends?
Sharon:
It’s not, but I was hoping that I can put it somewhere to make it available to other people to use it too because I do find it really useful. And I also created several other chatbots based on the structure of it. So yeah.
Larry:
I’d be happy to help you figure out a way to host it someplace and share it with folks. So we can talk about that offline. And if we do that, we’ll definitely share that on the site.
Sharon:
Yeah.
Larry:
But in one of the things you talked about, you talked about a lot of these… You’ve done a lot of these now, and one of the things you’ve discovered is that you need some kind of… As you develop all these sample materials and the scripted stuff that you do, you develop a repo, a place to put all of the content about that. Can you tell me a little bit about that repo or that database or whatever you call it, where you keep everything?
Sharon:
Yeah, yeah, absolutely. So also just want to be clear, this is not for training the large language model or data labeling stuff. This is just for the content that I create for the chatbot and also for the Cisco AI. I would say it looks really similar to any other content design repository that you would see elsewhere. The only difference being you have to really identify the different steps and actions behind those, I guess. Especially, if there’s a CTA, you really have to identify what’s the next step, what’s going to happen after you click on a CTA.
Sharon:
So I would say a lot of my time I actually spend on thinking about where this conversation would go instead of the actual writing, the content, the conversation of it.
And also, so this content conversation design repository, we call it Prompt Repository, is very much based on my previous content repository work. And I kind of tweaked it to make it more for this project that I’m creating for this ciscoai.com. And then I’m trying to build a similar one for the Cisco AI so that we can have a place for all the scripts and flows and also for all the texts, basically, for the AI system.
Larry:
Yeah. Now I’m wondering, I know a few people, several people in the conversation design world, and I wonder… As far as I can tell, that kind of stuff was always just stored in Voiceflow because that’s where you author and it ends up in there but it sounds like… Have you kind of teased it out of Voiceflow so that you can do other things with it? Or…
Sharon:
I think they do have a new feature. I don’t know if it’s launched yet, but I think they’re going to have a new feature where you can link Voiceflow with spreadsheet or Excel, where you can just pull all the content from Voiceflow or import it from a spreadsheet. But so far, I’m still doing it manually just because it’s easier for me to update all the content and I don’t want to make any mistakes. And I would always double check before I hand it off to the engineers or the design team for review. Yeah.
Larry:
Yeah, no I’d love to… We’ll definitely need to keep the conversation going around that because that’s a pet peeve of mine, is how poorly that kind of content is managed in many enterprises now.
Larry:
But I want to shift gears a little bit. We were talking… Not everybody is excited about this and is knowledgeable, and you’ve mentioned just all the resources you’ve consulted, all the learning you’ve done. I don’t want to shame anybody because there’s good reasons to be excited or not about this. But can you talk a little bit about how you work with varying… Because there’s probably people above you who are like, “You got to do more.” And then there’s probably people elsewhere who are just like, “I don’t want anything to do with this.” How do you work with all the variety of levels of enthusiasm?
Sharon:
Yeah, I think that’s a great question. Well, I’m still trying to figure out that too, but also at the same time, I realized that some of the reasons that people don’t really like about AI was because they don’t really understand what AI can really do or cannot do. So I feel like one thing that I did in the past was that I would always tell people that AI is not for everything. It is not as powerful as you think. And I would also like to share some of the resources just because I feel like nowadays a lot of people are working on AI or being pushed on AI project without knowing anything about AI.
Sharon:
So I would always share my resources with them to help them to learn something about AI at least. And also, a lot of times, especially when I’m creating this content design style guide, chatbot, I would always think about what other people might use this tool, how they’re going to use it in different use cases.
Sharon:
And that’s also why I wanted to try to keep this chatbot as simple as possible. I don’t want to over complicate it. And I wanted to just provide guidelines and recommendations instead of trying to write the thing for users so that people won’t feel like their job’s going to be replaced or something. But it’s more for helping them to speed up the process to make their life easier and to make things simpler, I guess.
Sharon:
Well, I think for other people who are already really knows a lot about AI, I think it’s… One way to do it is I like to talk to people who know more than me. So I would always try to see if people have any recommendations around learning materials or even any good use cases for AI. I would always ask people this question: what’s the best AI products that you have ever experienced out there? And I would try to see what’s the good thing about those products.
Larry:
Nice. So you’re just staying curious and staying out there. Hey, and that reminds me, I want to go back a little bit to that… you’re skilling up to this. That sort of points… one of the main ways you’re getting up to speed on the technical stuff. But you said something before we went on the air that if I’m remembering this right, that sometimes some of the technical feasibility, that in a traditional content design workflow, you’d probably talk with an engineer about it. In your current role, sometimes you have to be the person making the decisions about technical feasibility. Can you talk a little bit about how you got prepared to do that?
Sharon:
Yeah. So I think this is also really interesting. So I was put on this project, this Cisco AI projects really, really early on, probably at the very beginning of the project. My manager at that time was super supportive of me. I told her that I wanted to work on AI. That was before the Cisco AI project. And I told her that I wanted to take courses and I wanted to get a budget for me to be able to learn AI and get trainings and stuff like that. And she was really, really supportive. And also, she put me on this project. The first…
Sharon:
She put me on this project and she asked me if I wanted to be the content designer of this project. And she really, really trusts me on this. So that’s how I got started on this project. And then, because by the time I started that project, I already have learned a lot about AI, so I was the subject mat ter expertise on that project. And I would say this is really important just because this is another way to show people how content can influence our product. And we didn’t really have a lot of chance to do that in the past.
Sharon:
So I feel like it is just one of those things that can add a lot of value to content designers. And also because I think most of the AI product right now are still very much content-based, and people are realizing that they need a content folk on those projects so that… to help them, not just to create the content for the interface, but also to build the logic behind it and to think of everything from a content perspective. So, yeah.
Larry:
Yeah. No, I love that. And just by coincidence, a scheduling coincidence, this episode will air right after my interview with Andrew Stein, and I know he was your mentor at ADPList. And then you did almost a switcheroo that… You’re mentoring each other now, which I love that. But that’s one of the things he talked about is that… he talks about a lot is that by cultivating all of this, you’re seen as an expert right off the bat. And as you’ve demonstrated in just a year, you’ve become an internal expert on AI, so that’s…
Larry:
Hey, but Sharon, I can’t believe that we’re coming up close to time already. These conversations always go way too fast. But before we wrap up, is there anything last, anything you want to revisit from the conversation or just make sure that you share before we wrap up?
Sharon:
Oh, yeah. I wanted to recommend a book. It’s called Co-Intelligence, I think, for anyone who’s interested in AI. I feel like you don’t really have to spend a lot of money on those courses. Because I feel like things are changing so fast nowadays. It’s possible that tomorrow there’s going to be a new AI tool that everyone’s going to start using it. So if you wanted to learn the history of AI, what it looks like before the GenAI, the GPT age, I think… And also different use cases of AI and what you can do to use AI as a system or use AI more like… I don’t know, to actually do things for you, on behalf of you, it’s a really great book about AI. Everything you wanted to know about AI is in there.
Sharon:
And I would also say just do it. Just start building things. Just start playing around with different AI tools, AI products, and to talk to people who build those products are really, really important. And start making connections. And also, welcome to talk to me if you’re interested in AI and conversation design. I think there’s definitely going to be new paradigm in the next few months or years for a conversation desire, even for content.
Sharon:
I think AI is going to completely change the way we work. And also there’s going to be more personalized content than ever. Right now, I think content is the data and content will help us to understand our users in a much, much deeper way than we ever were. So just be mindful of those. And also, yeah, keep learning, I guess.
Larry:
Yeah. Well, I got to say, this comes up over and over again in these conversations that content is the engine, that is the thing that makes all these things run. And we are the content people, so… And you’ve demonstrated how much you can do with that. Hey, one very last thing, and you just mentioned folks staying in touch. If folks want to connect with you or follow you online, what’s the best place to find you?
Sharon:
Oh, LinkedIn, definitely.
Larry:
Okay, great. Well, I’ll link to that in the show notes as well. Well, thank you so much, Sharon. I really enjoyed the conversation.
Sharon:
Yeah, same. Thank you so much. Bye.
