Decoding AI Product Marketing with Abhishek Ratna

Sheikh Shuvo:
Hi, everyone. Welcome to Humans of AI. I'm Sheikh. And this podcast is where we learn about the people who are behind the incredible tech that's changing our world today. I'm joined by Abhishek Ratna, who's Director of Product Marketing at LabelBox, an industry-leading data AI platform. Thank you so much for joining us.

Abhishek Ratna:
Great to be here, Sheikh. And thank you for inviting me.

Sheikh Shuvo:
Yeah, absolutely. Now, you've done a lot of cool things in your career and a lot of complex things, but if you had to describe your job and what you focus on day to day to a five-year-old, what would you say?

Abhishek Ratna:
I would say I help people understand how my company solves their problems. And to make my company sound really popular and cool, I'll share how my company makes heroes out of people who are using our products. I tell these stories through websites, videos, articles, and more things. I'm basically a grown-up storyteller.

Sheikh Shuvo:
Yeah, that sounds like a great business card right there.

Before we dive into what LabelBox does, can you tell us a bit more about your career story and how you landed where you are?

Abhishek Ratna:
Absolutely. So I started my career, my first job out of engineering school was as a developer for four years in India, but I used to test web applications for big banks from secure locations within India. My background kind of lay in growing up with a very healthy dose of stories and storytelling. I did a lot of volunteering with a wellness and meditation NGO in my teenage years, and I continue to actually do that to this day. So, I felt my calling lay more in working at the intersection of technology and people. And that's when I decided I wanted to switch careers.

So about 15 years ago, I enrolled in an MBA program at the University of Florida, where I specialized in a degree in MBA in Marketing and Strategy. And for the last 14 years, I've been a marketer in different capacities. Now, I've marketed everything from business applications to women's and kids' fashion, to developer SDKs, to cloud-based data engines, to ad products, to AI tools.

My experience has spanned demand generation, growth marketing, developer marketing, and product marketing. So that's my background now.

Sheikh Shuvo:
Within your wide range of industry experience in various product marketing roles, would you say there's anything unique about the challenges of product marketing for AI products versus other things in the tech world?

Abhishek Ratna:
It's a very good question. And I think I'll start by saying that at its core, fundamental at its most basic level, marketing an AI product is very similar to marketing any technology product. The goal is to connect the product company to people's expectations and needs. That's a simplistic definition. Now, the thing is, what's fundamentally different is people's expectations and needs around AI are different from traditional products. I've talked about this in previous sessions and other talks that I've done in the past, where AI products tend to embody four principles more so than other products.

The first of them is automation. We just expect AI systems to be more intelligent, more cognizant, and more capable of automatically understanding our requests and creating very detailed and powerful responses that address our needs. There's always that question of how does your AI product or solution help automate my problems or help me create more automated products.

The other aspect is intelligence. This is a vague concept, but there's a lot of industry momentum behind the idea of intelligent workflows and intelligent applications. It ties in with automation where the software or the product understands user needs in a much deeper, more nuanced, and contextual way. So, bringing out that element of intelligence is key to differentiating your AI-led products.

And the third pillar is ambience. Consumers expect AI to seamlessly work from within the context of the customer, where the customer is. They expect AI to work out of a smartwatch, a phone, a laptop, an intelligent device at home. We are expecting not to sit in front of a computer or be tethered to certain devices to use AI. We want it to feel very natural and like a seamless extension of our lives. So when positioning AI products, that's another area of focus.

So I think those are the three pillars of messaging or positioning AI products that I think are significantly different from regular products.

Sheikh Shuvo:
Yeah. That's a great framework. To use looking at your own workflows and day to day, are there any things that have changed or ways that you've incorporated AI tools to accomplish what you do?

Abhishek Ratna:
Oh, yeah. I'm a big believer in using AI. I've actually done a few articles and talks also on how I use AI within my product marketing or marketing workflows. And I feel, I think we're in phase one right now. Like, you know, ChatGPT showed the world that there's so much possibility of like innovating with AI. The story is, I do believe like the technology is not, will definitely take more time to mature where it's at a level where it's kind of meeting or passing human intelligence, especially on more demanding tasks. But I feel there's a huge leg up that we knowledge workers and professionals can get out of AI even now. The way I've used it is like on three, three or four things. The first of them is ideation. It's a very easy, like both in product marketing and demand gen. Like I found writer's block to be a real thing when you're trying to find that perfect pitch or that perfect angle to position your products to customers.

So how do I map my product's capabilities with my customer stated problems and needs with like how the industry is talking about a product? With AI, I found it very easy to give, I use a bunch of AI tools, but I also really love using cloud by Anthropic. That's one of my favorite go-to models. So one of the things I would do is I'd kind of curate these snippets of text, which I think are very relevant for coming up with content ideas and then feed them into these large language models and ask to create five or seven ideas for me. And it's been surprisingly good. I've seen these LLMs come back with very interesting angles or pitches which are not too far off which a little bit of refinement allow me to come up with that perfectly positioned perfect positioning or pitch and that would have taken me that could have taken me weeks in the past, but now I can do that like over 30 minutes.

So being a huge lift for me, another way in which I see AI, I there's more automation, like if it's a non-business, non-critical piece of message or content that I need to produce, actually trust AI even more to like, to help draft or almost in some cases, even create the finest output for some. Service messages, in-product messages, or notifications, like those kind of customer-facing comms can be almost completely automated with AI in my experience.

And I've been a, I've been toying with ways to make those workflows even more efficient. And the third way in which I use AI is like as a research assistant. And for that, I really love Perplexity AI. That's my go-to AI search tool. I've used that for a lot of AI-enabled search. I've used all of these, like I use Bing, Bing AI.

I'm also signed up on the Google experimental feature, so I get access to Bard-enabled searches. But perplexity has been my favorite, like in its sophistication and its ability to understand very complex questions and give very on-point responses. So I'm a big believer in perplexity and I've used that a lot to understand complex industry problems or understand complex features or capabilities that I need to talk about.

And it's been really good in helping me understand stuff. I think the final way in which I use, I've used AI is as an analysis assistant, where I've kind of given AI tools, a bunch of unstructured text. And I've asked it to create tables and pull out and categorize comments or content. Or like just create tables of data from unstructured data or like, or take an existing table and then you find patterns in that data.

And I found out of the box performance from both cloud and ChatGPT to be super, super high quality those ways. So those are the four ways in which I've used it, like from core content creation to automation of content production, to doing more analysis, to doing research, I've used it for all those four things.

Sheikh Shuvo:
Absolutely. It feels like having a small army of interns always available to help. Yes. Awesome. Absolutely. Well, given those, tell us more about what LabelBox does and how you're working with different enterprises.

Abhishek Ratna:
Absolutely. So LabelBox is a platform for doing data-centric AI. And what that means is data is the lifeblood of any successful AI or ML product, company initiative, and LabelBox has been used by many Fortune 500 and Fortune 100 organizations to build very high-quality datasets quickly for making sure that their AI models, their AI products create outputs that are highly accurate, that are trustworthy, and that are of very high quality. Now, Labelbox has evolved quite a bit. The company was founded about five years ago, in 2018. Our strength has been that customers love how LabelBox created intelligent workflows and customizable workflows, created a platform for managing large data labeling teams. And then LabelBox went ahead and now we have our own team of experts worldwide who are really knowledgeable about understanding the nuances of labeling data for machine learning.

And so we pair this automated platform with this always on, always available human labeling team to help companies solve for their data annotation problems. But what's interesting is, LabelBox has evolved quite a bit from just annotation. There are a lot of ways in which some of our key customers actually use LabelBox directly to power AI-enabled searches, or they use LabelBox to analyze, without any human intervention, millions of structured data pieces, images, photos, videos, texts, and understand patterns, insights in minutes or seconds, way faster than what any complex engineering workflow would require. So the way I would describe LabelBox's capabilities today are like for building AI systems and building models, LabelBox allows you to build the most accurate datasets much faster through a combination of automation and world-class workflows. It also allows for things like evaluating models and especially with transparency, allows you to evaluate if the outcomes are safe, trustworthy, and highly accurate. And then the other ways in which customers are finding value out of LabelBox is in using AI, like you have tons of data sitting in your data lake or in your warehouse, and you want to automate data enrichment by applying foundation models. You want these large language models to automatically go and tag up your data, classify or categorize them in some ways. LabelBox helps you with that too. And I've seen a lot of customers use it for that. So that was my wordy definition of LabelBox.

Sheikh Shuvo:
Sounds like a powerful tool. I'll definitely add some links that the listeners can look at to the content you've made around LabelBox. It sounds like a great tool. Shifting gears a bit, going back to your own career journey, you've worked at some of the biggest tech companies in the world, Microsoft, Google, Databricks, and now you're at a smaller startup. What have been some of the things that you've had to adjust to going from a bigger corporation to the startup world?

Abhishek Ratna:
Yeah, that's a great question. Let me start by what's common. All these companies have a lot of problems and there's a lot of gray area. A common misconception is that people think big companies have it all figured out. The companies you mentioned, Microsoft, Google, Databricks, I also worked at Facebook, and all of them, in the roles I've had, there's been a lot of white space, a lot of room for analysis, for building structure, and for bringing more rigor to the way we did things. Even in the largest companies, you're always building. It's often a function of which team you work for. I've seen more changes or variation inside of a company within different teams than across companies sometimes.

Coming to startups, the biggest points of difference start with the obvious ones. Startups are evolving, so priorities change very fast. There's a premium to execute; timelines are shorter. There's a need to be more decisive, adjust on the fly, innovate, and work fast. Startups pay a premium on these aspects. One of the pros of working for a startup is that all the work I've done directly touches customers. In larger organizations, it could be more rounds of review and internal validation before the work finds customers, but here the results are immediate and tangible.

The third thing with startups is embracing the mindset that you're here to build the process as you execute. There are high pressures to produce outcomes, but processes can be lightly defined, and there's always room for improvement and growth in the maturity of operations.

The fourth biggest pro of working for a startup is being embedded into the business in a way you often don't see with larger companies. The access to founders, the transparency, and access to business decisions that define the survival or course of your business are more prevalent in startups than in larger companies, where you can sometimes feel like a part of a machine with a very limited scope. That's how I would describe the startup life.

Sheikh Shuvo:
That's finding joy in chaos and responsibility.

Abhishek Ratna:
Very well summarized. Yep.

Sheikh Shuvo:
Yeah. Well, the very last question I have for you, Abhishek, is for someone who's just graduating college right now and is interested in getting involved in different product marketing roles, what are some of the tips you would give that person as they evaluate different companies? What are the things a new product marketer should be looking forward to evaluate a role?

Abhishek Ratna:
I think because so much of our roles in product marketing comes down to telling effective stories, looking for customer centricity is important. A product marketer should develop an eye for product-market fit and try to understand, get a 360-degree view of whether the company's capabilities and stated features align with real customer problems, and also vice versa, what's the customer feedback on that company's product. Look out for resonance and coherence between the company's messaging and what the customers are saying about the company. Understanding the roadmap, the direction, and developing an inquisitive mindset around whether that roadmap aligns with how customers or the industry is evolving are also key. It would be your job often to define these things to the customer. If the founders really understand the industry, if the team really understands the industry, if the products are on point with customer expectations, and if there's a well-oiled machinery between customer feedback and product development, then your job becomes much easier. Otherwise, you might find yourself in a situation where you're creating positioning and messaging that's not landing in the market, and then you might be left wondering what went wrong.

Sheikh Shuvo:
Sage advice. Awesome. Well, Abhishek, this has been a super fun conversation with lots of actionable tips for listeners to take advantage of. Thanks again for joining.

Abhishek Ratna:
Thank you. This has been great. Yeah.