Blending AI & Human Expertise For Better Content Creation

Tessa Burg: Hello, and welcome to another episode of “Leader Generation,” brought to you by Mod Op. I’m your host, Tessa Burg, and today, I am joined by Shaheen Samavati. She’s the co-founder of VeraContent and an expert in content, creation, development, and distribution in a global environment. Shaheen, thanks so much for joining us today.

Shaheen Samavati: Thanks for having me. It’s great to be here.

Tessa Burg: So we have been spending a lot of time on this podcast just talking about how different areas of thought leadership, expertise, and skills in marketing are evolving as a result of AI being introduced to our lives and just in general how we get work done. What have you seen overall as the impact of AI in content development?

Shaheen Samavati: Yeah, so clearly this is something that’s been at the top of my mind being someone who is, I’m a founder and CEO of a content creation agency. So VeraContent, we basically do multilingual content creation projects for global brands and our bread and butter is creating the content for these large, ongoing projects. We do blogs, we do social media projects, so we’re creating large amounts of content and we’re using human teams to do that, at least historically. But I mean, basically over the past several years, there’s been that introduction of AI tools out there even before ChatGPT came onto the scene. So it was something that we were looking into and thinking about how we might be adopt these things because we’re always looking at how to use automation in our processes. And I’d say even before ChatGPT, we were using some AI tools for things like transcriptions and things like that, and then now with I’d say the vast improvement in the quality of content generation capabilities of these large language models, we’ve really had to think about okay, now, and also just the publicity around it when ChatGPT came out, that every client was expecting okay, how are you gonna use this to speed up your processes? And we’re like well, we’re already being very efficient in our processes. We’re using these, these, these tools, but okay, now we have to also see how we can use ChatGPT specifically so. And of course, we also, so we spent a lot of time, we actually created a task force in our company to analyze how we could use ChatGPT or other similar large language models, tools that use them in our processes. And so we, we tested more than 20 tools so, and yeah, and we’ve developed some best practices within our company now and how we’re using them. But yeah, what was the original question? I went on a tangent there.

Tessa Burg: Oh, no, no, I’m glad you gave us your background because it definitely takes adaptable leaders to embrace using AI and sort of go beyond just accepting that it’s here and it’s gonna change the way we work, but use it in a way that does elevate the work itself. Elevates creativity, elevates accuracy and distribution, and you are the perfect leader to do that because you were originally from the United States, but now your company, as a global content agency, is located in Spain. Tell us a little bit more about that journey and where the concept for VeraContent came from.

Shaheen Samavati: Yeah, so my background is as a journalist, so I’m actually from Cleveland area, Akron, Ohio, but I worked at the Cleveland “Plain Dealer” and other newspapers, and basically that background in newspaper journalism I think has really influenced the vision that I’ve had for VeraContent. But in between, I did an MBA in Spain, that’s why I originally went there. And after that I worked in communications and marketing and I led marketing at a scale-up company. So basically I had to create large amounts of content and doing that within a company, and seeing the challenges that were around that, and how it was difficult to find partners who could help with that basically and I saw the opportunity there. So that’s why I originally started VeraContent, especially in Europe because there’s so much need for multilingual content. Obviously a lot of new demand for English content, so that’s kind of how I got into it. Being an American in Europe, everyone wants English marketing content to reach global audiences. So I started working with European scale-up and larger companies who were growing their content operations and helping them do that, and then I saw that there’s demand for more languages. So basically I founded the company with two business partners. We started building out our capacity to work in more languages and it’s really like a chicken and egg thing where you really have to have the team in place to be able to ensure the quality, but you have to be able to judge quality to make the right team so it’s it takes time to build that capacity basically, and that’s what we’ve been doing there so.

Tessa Burg: That’s awesome, so tell me a little bit more about building the team. So you do have so many different languages you have to accommodate. Has AI played any role in giving the team maybe some more flexibility or capacity as you’re trying to serve so many diverse target audiences?

Shaheen Samavati: Yeah, so our expertise definitely is having those human linguists who are experts in the language and experts in marketing in their home markets, so we have people who are based in each of the markets, who know the market, who are up-to-date on local trends. And I think that AI has, I mean, as I was saying, it’s been since the beginning of the company, AI in some capacity has been around for a while. So we’ve been using things like AI transcription for a really long time and that’s something that allows us to repurpose content. For example, we do an interview, we can transcribe that and then we can use that to create different pieces of content out of it. I’m sure as a podcast host, you probably use some AI transcription for getting your, I don’t know, it’s common a common tool now. But so, and things like adding subtitles. I mean, obviously with translation itself, localization. I mean, we’ve been using a CAT tool that has some AI features built into it and that they’ve been growing over time. So really it was AI goes a lot further than only the generative AI I guess function of tools like ChatGPT. But of course, now that that has advanced so much because that’s something that we had even been looking into before ChatGPT. There was the GPT language model available in the API that a lot of other tools were using and there’s things like Jasper and Writer and other tools that we had been testing. But really three years ago, they just were this is like gonna take a long time before it can ever do anything compared to a human and then came the next versions of the GPT language model. By the time we got to GPT-4, it was quite impressive and it’s like okay, we really have to take this seriously and see how we can incorporate into our processes. And I think it’s something that for a, I see how it’s amazing for entrepreneurs or someone who is creating content. They’re kind of the thought leader and they just wanna proliferate their ideas, and I think that that these kind of tools can help you. If you have the content and the basis, it’s really easy to create lots of content from that, but you need something quality to put in in order to get the results out. And I think it’s difficult to scale that. So when you’re a larger company, lots of companies, our clients are coming to us and saying okay, I wanna do more and more content using AI, and it’s not so simple to just push a button and get a lot more content out of it because the results you’re gonna get from that aren’t probably gonna meet your expectations. So I think it’s a lot about creating those processes to make sure that you’re consistently getting good results out of the AI and for us, that’s been creating a lot of guidelines for how we use AI, when we use it, in what circumstances, and I think that’s the challenge. When you’re working with a team, it’s a whole nother level than if you’re just one person who wants to create content yourself, right? So that’s that.

Tessa Burg: Yeah, no, I definitely can relate to that challenge. We, as an industry, I think spent the last year testing lots of different apps and getting really comfortable with ChatGPT as a part of the process. And we’ve seen a lot of benefits including saving time, boosting creativity, but then scaling it and actually having any one tool be enterprise-wide for the purpose of creating quality and elevating the experience for your clients is a different challenge. It’s not enough to be like oh, we love this tool. Let’s just go ahead and use it or just have everyone use it. You have to have that process, those standards, those quality checks in place first.

Shaheen Samavati: Yeah, absolutely, and for us, it’s been with single client project seeing whether it makes sense to use the tool in some way or not and for each project, creating guidelines, creating custom prompts that’s like okay, if you’re going to use this prompt, don’t do this way, do it this way. So it’s a lot of, yeah, a lot of documentation and things that going into I think using the tool well.

Tessa Burg: Yeah, so you have to really believe that the quality’s gonna be higher because while there’s some time savings, it ultimately gets invested into a different area like time.

Shaheen Samavati: Yeah, absolutely, and actually we’ve done a lot of testing on that just because clearly the goal is efficiency and it’s like can we speed up? And for some types of content creation, it’s not very helpful at all. When we’re writing something that’s a really unique piece of content that it needs to be maybe, yeah, that has a specific format that it’s not formulaic. I would say when we’re doing, for example, we have some projects that it’s we’re always gonna do a certain type of article in this format and so we can create a custom GPT that says okay, input these things and then get a outline of the article, and then we can use that as a basis that speeds up things for the content creator. But then there’s other things that are not so formulaic that it’s just not that useful to use the tool and I think, yeah, it really, yeah, it really depends on the particular project so.

Tessa Burg: I think a lot of content requires actual human deep expertise in specific areas even if you’re just checking the content. And our use of ChatGPT so far, when we use the publicly available version, is that all the insight that it brings back is solidly mediocre and if you are looking at it to improve quality so that you can justify these additional processes, these skill trainings, what you’re investing in the tools themselves, it’s really important that you don’t lose that person expertise in the process.

Shaheen Samavati: Yeah, that’s what I was actually getting at. The results you get out of it are only as good as what you put in, so it’s really important that you’re putting in some unique insights in the beginning if you wanna get an article that includes those insights because it’s obviously not gonna come up with those things. It’s only really good for, I mean, it can be good for brainstorming and thinking about things in a way that maybe you you didn’t like asking it for a bunch of different potential titles for something, it might jog an idea in your head. In my experience, I almost never use the options that it gives me, but it might help me think of something good. And then the other is formatting because a lot of times you have content in one format and it really can help you put it in another. So for example, yeah, you have a blog post. You want a social media post about it. It can generate that for you and then you can rewrite it or have a writer on your team rewrite it. But I think repurposing content I’d say is our number one use. We’re almost never relying on it to actually create the content.

Tessa Burg: So you can save time in sort of research, brainstorming ideas, summarizing transcripts, repurposing content. But what you’re finding is don’t lose the expertise in the industry, in the subject matter, or in the localization expertise for different cultures and different reasons to engage in content. Another area that we’ve seen ChatGPT and really generative AI have an impact is how people interact with content in general. Whenever we’ve done B2B content campaigns, we always have an element of SEO baked in. We wanna be cognizant or we’re very aware that 80% of the content is discovered through Google. Have you found that people going out to ChatGPT to get advice, to learn more has impacted the way content is discovered and consumed?

Shaheen Samavati: I mean, yes, absolutely. It seems to be going that way. I mean, even Google now has its AI summary that you can can see. So I think that yeah, everything’s going in the direction that people are using these chat tools to search. I think the good news is that the criteria for what it shows you, what come up as the top results tend seem to be similar to what search engines use. So for example, if you have really well-optimized blog posts or content on your website and that’s been ranking well in Google, chances are if you ask a question related to that to ChatGPT, there’s a high chance you would show up in the results if you are among the top results in Google. So it does seem that best practice of SEO does seem to help for ranking in ChatGPT results as well.

Tessa Burg: Has anyone expressed any concern around loss of traffic since ChatGPT doesn’t have links back to websites?

Shaheen Samavati: Hmm, yeah, it’s a good question and it’s not something that we’ve heard a lot of concern from our clients yet, but yeah, it’s definitely, I mean, actually ChatGPT does give links in some cases. I was actually testing the other day, searching for multilingual content agencies and seeing if it showed up, and it didn’t link to my website but it linked to a blog post that recommended us. So there can be like a link chain from ChatGPT. So it kind of gave me, I asked for the top agencies, it gave me its top 10 and then it gave me a link kind of justifying as a source for each one.

Tessa Burg: So maybe more research and discovery type questions. Almost early customer journey, probably not so broad. But more bottom funnel types of queries, it will give you exact answers. Yeah, we also have not had anyone start seeing major traffic drops and I think that’s probably just indicative of people still using Google a lot for when they need really some more bottom funnel queries or to make sure that whatever they got in ChatGPT was accurate. I found myself doing that. I’ll brainstorm and get ideas from ChatGPT, but then I’ll be like okay, let me just make sure this is real because we’re so aware of hallucinations, that’s cool. And I have not found the Google AI summaries to be helpful at all, so I feel like ultimately maybe it’ll evolve there but.

Shaheen Samavati: I don’t trust the Google AI summaries at all. Half the time it tells me is something totally wrong. So I trust that search results a lot more than that, yeah.

Tessa Burg: Yes, no, I agree. So what are some other considerations when we’re thinking about the way that we’re creating content is different, the way people are interacting or finding content is different, and when we expand this out to a global environment where there are different standards like localization is different around the world and especially in Europe, what are other considerations that content developers have to take into account?

Shaheen Samavati: Yeah, I mean, I think it’s really exciting that it’s enabling more communication across different languages. I feel like if you have a general notion of another language, I think these large language models can really help you communicate better in that other language. I think it is dangerous when you have no knowledge of the other language to rely on those tools to create content in that language for you because obviously they’re not perfect and I don’t know if they ever will be. So at least for now, it’s super important to have a human involved in that process who really understands the message that needs to get across I think, yeah. I know there’s also these tools for language dubbing, so you can make a video of yourself or take any marketing video you might have and run it through this tool, and it’ll create audio voiceover using AI that even sounds like the same person’s voice and it’s in the other language, but actually some of these tools allow you to edit. So I mean, I think they all will eventually. There’s one called Rask that is really cool that has, basically similar to a CAT tool within it. It does the translation for you and then you have the translation side-by-side and then you can go in and edit the transcription to make sure that it’s accurate. And I think I would never put out a marketing video with an AI voiceover for any reputable brand without having someone really check that. Be very embarrassing if you didn’t so.

Tessa Burg: Yeah, that is very cool. I mean, one of the areas that we focus in is a lot of B2B marketing, a lot of manufacturing, energy, transportation, skilled labor. And right now in those sectors, knowledge transfer, training, and upskilling is huge and it is for sure going across multiple languages. So I love this idea of using something like Rask to help automate that and then make sure you’re not losing the person expertise to go back and edit. But still it saves a ton of time. I mean, as someone who’s been in these times. And I think we’re at this unique point, not just with artificial intelligence, but that there’s so many different generations in the workplace and how important it is to make sure that we’re not losing that knowledge and expertise from people who have been at the company for 25, 30, some 40 years. And just like we’re talking about with content development, there’s so much humanness that you don’t wanna lose from those jobs and capturing transcriptions, capturing audio interviews is a great way to ensure that you don’t lose that knowledge and skill.

Shaheen Samavati: And actually one challenge with localizing audio visual content is that having to line up the timings of everything and that’s something that AI tools really help with. This tool, Rask, for example, it really puts side by side and it automatically adjusts the lengths of, using AI, it adjusts the phrasing to match the same amount of space that you have in time so it matches up because when you’re doing a dub, sometimes in Spanish it takes way more words to say something than in English, for example, or any language. But I find that using the AI tool to do that, even if you are planning to do, a lot of brands don’t wanna use AI voices. They’re not ready for that. They’re not convincing enough. It’s the uncanny value thing. So but I think it’s still a good idea to use a tool like that to speed up the process of matching up and making sure that the language lines up, like the timings, and then you can have a human record it afterwards so.

Tessa Burg: I love that solution. So I took a note of that tool. We love discovering and testing new tools. What are some of your other favorite tools that you’re using in your process?

Shaheen Samavati: Yeah, I would say, well, besides Rask, we also use for audio visual stuff, Subly is the tool we really like because it works in a lot of languages and it’s really quick for adding subtitles to videos. In that same vein, there’s a lot of video tools the IO Descript and we actually are using, depending on the project because there’s so many tools that have one capability and one has one and one doesn’t have the other. So yeah, I won’t go into the details of all of the different functionalities, but we found, for example, Descript has the option to do inline editing. You can mark out text and it automatically moves it from the video, and that’s really useful for talking head videos. But then we have other editing tools like V that are more useful for creating social media type videos, so there’s that. And then when it comes to text content creation, I’d say our favorite tool and the thing we use the most is Grammarly which of course, was not seen as an AI tool, but they’ve added in a lot of AI features and it’s really powerful, the style guide features that it has. Of course, Grammarly only works in English, so we use Grammarly for English projects and we have another tool that we use for other languages which is called LanguageTool that has some of the functionality of Grammarly and it also has the ability to create style guides. And for us working on projects when there’s so many different writers involved, having consistency and making sure everyone’s using the same terminology, characterizing the brand the same way and everything, being able to create rules around that and you can basically have, when you have a kind of premium company account on either of these, you can have things flagged. While you’re working in a Google Doc, it says hey, that’s not how you write that basically. So creating your own custom rules and then I mean, using custom GPTs in ChatGPT is also something we’ve been implementing and in terms of generative AI for writing, that’s the tool we’re using right now even though we’ve investigated tons of them. But we found that that custom GPT feature has been really useful to be able to basically share pre-made prompts across the team using that.

Tessa Burg: Oh, I was gonna ask what are the custom GPTs doing? That is a great idea. We started using the OpenAI API to create brand agents that are specific to the knowledge, brand guidelines, style of that brand for a very similar reason and I’d love the idea of connecting that with localization tools, with a Grammarly or LanguageTool, ’cause if that has an API, then you can get in those localization standards as well for when you are starting to distribute that campaign globally or even just generate ideas, and make sure that it stays within the voice of the clients in that market.

Shaheen Samavati: Yeah, absolutely, a lot of localization software is also adding in AI features. We use one called Phrase. That’s what we use for translation management and localization management and I mean, they’ve added in this, I forget what the name of it is, but basically it will suggest ways to rephrase things and stuff within the translation tool which is just really cool. And I think it’s only going in that direction If you’re not already proactively adopting AI, I think everyone is gonna end up adopting a lot of bit because so many of the tools we’re already using are just integrating more and more AI features. So that’s what we found.

Tessa Burg: Yeah, no, we have found the same thing. When we started, man, I feel like was it 2018 is when we really started using AI-enabled tools as part of our process and we called it the AI tech stack. These are the things that are AI and we’ll build some of our own solutions, and now it really is just the tech stack and you have to be aware of where AI is being used and the data that’s going in and out, quality standards, those conversations. It’s more we have to look at our whole tech stack as AI-enabled and apply the same sort of caution that we’re using in ChatGPT to everything we do, but we have found that the quality is better, so it’s worth, again, we we’re at a point where we’re starting to scale it enterprise-wide so we’re having the upscale and the training. But until everyone has seamlessly integrated this and we have custom GPTs or for us, we’re building a library of agents to automate checking quality standards, I’m not sure we’re gonna see those significant time savings that you get as an individual. I feel like on an individual basis, you’re like oh, my gosh, this is how this works. And if you’re a three-four person agency like you said before, really small companies, entrepreneurs, they probably see the time savings immediately. But then when you move to enterprise level, doing things for multiple markets, over three markets, it requires a level of discipline that, it requires more time investment to get it stood up.

Shaheen Samavati: I find too why it’s so advantageous for an individual, someone who’s not a writer and needs help writing, it can help you with that. But in our case, we’re all professional writers. Everyone’s very efficient at their jobs already. The people who write for us, that’s all they do all day long is write. So having this tool doesn’t always speed up their time that much basically. I would say we have been comparing how long it takes us to write an article with and without the tool and I would say it’s not a huge significant time savings, but the results are better and the level of consistency is better because of, like I said, creating these rules, making sure that we’re using the right tone of voice, using the right terminology. Yeah, and we’re always being consistent in the format and all these things. We don’t make maybe human mistakes that we would’ve made in terms of the the output, but I feel the actual time it takes to create the content has not been shaved off very much in our case.

Tessa Burg: Yeah, and I feel AI, everyone getting more comfortable realizing what it can do and what it can’t do is actually elevating the value of expertise as much as it is data. I think people are realizing oh, my gosh, the AI is only as good as the data that comes in. Well, it’s also only as good as the person using it and their expertise and experience within that field, within that culture, within that industry. Or else you will not get the true lift in performance or the true lift in what it can do creatively and creating new experiences that really resonate and are personalized to the consumer, to your target audience.

Shaheen Samavati: Yeah, absolutely, I agree.

Tessa Burg: So Shaheen, this has been a great conversation and I’ve learned a lot. I’ve jotted down a lot of tools I’d love to test. If other people wanna reach out to you and learn more, what’s the best way to reach you?

Shaheen Samavati: Yeah, definitely our website, veracontent.com or LinkedIn, Shaheen Samavati.

Tessa Burg: Perfect, and if you want to hear other episodes of “Leader Generation,” you can visit ModOp.com, that’s M-O-D-O-P.com. We’ll have the notes from this interview on the website, as well as all the tools mentioned in case you wanna explore for yourself. And Shaheen, I hope we get another chance to talk soon. but until then, thanks so much for being our guest.

Shaheen Samavati: Thanks for having me, it’s a pleasure.