Marketing and sales in the age of AI
Gian Dela Rama is the chief product and artificial intelligence (AI) officer at Sprout Solutions, where he leads product management, AI research, machine learning engineering and data science at the largest business-to-business software-as-a-service (Saas) company in the Philippines.
With over 21 years of experience, his leadership has driven major innovations, including the relaunch of ReadyCash and the development of ReadyWage, both of which have accelerated Sprout’s growth. He also founded Aiah.Ai, one of the first AI startups in the Philippines, which developed KIRA, the country’s official COVID-19 chatbot.
Question: Could you explain what ‘agentic AI’ means, and what sets it apart from traditional AI applications in sales and marketing?
Answer: Agentic AI refers to AI systems that can operate autonomously, make decisions and pursue goals with minimal human input. What sets it apart is its ability to reason and act, not just respond.
To understand what makes agentic AI special, it helps to first look at the more familiar forms of AI. Discriminative AI is the kind we see in systems that recognize patterns and classify data, like email spam filters or fraud detection in banking. Then there’s generative AI (gen AI), which powers tools like ChatGPT and Claude. These models create new content based on what they’ve learned.
Agentic AI builds on those capabilities. It uses enhanced versions of the same large language models behind gen AI but adds a layer of reasoning that allows it to break down complex problems and figure out what steps to take to reach a goal. It’s that ability to plan and act autonomously that sets it apart.
In the world of sales and marketing, we’ve seen discriminative AI used for things like social media listening or customer segmentation, while gen AI shows up in lead-generation chatbots that can respond to customer questions.
But agentic AI goes further. Imagine inputting an ideal customer profile, just that one prompt. An agentic system could convert that profile into specific industry descriptors, identify decision-maker roles, query LinkedIn and other databases, compile a list of prospects and then rank those companies based on size or number of relevant contacts. All of that without human intervention.
Q: What advice would you give to sales leaders who are hesitant about adopting AI, especially those in traditional industries or with less technical experience?
A: The great thing about today’s consumer AI products is that they’re built for nontechnical users. The learning curve is low, and all you need is the ability to type into a chat box. My advice is simple: best way to start is to dive in and experiment. Pick a chatbot you’re comfortable with—ChatGPT, Gemini, whatever feels natural—and start asking it questions related to your day-to-day. Try something like, “Write an email pitch for a COO (chief operating officer) based on our company website,” and see what comes back. Once you’re comfortable, start pushing the boundaries, ask it to do more complex tasks, solve bigger problems.
Along the way, there are three things worth remembering. First, even with how far AI has come, it’s still prone to hallucinations; so always double-check the facts. Second, never input sensitive or proprietary information. And third, give yourself permission to explore. The more open you are to discovering what it can do, the more valuable it becomes.
Q: How can businesses gauge whether their market is ready for AI-powered products?
A: It starts with understanding the customer and the use case. Too often, product teams build based on their own assumptions about the problem, rather than what the customer actually needs. That can result in solutions that feel innovative on the surface but fail to drive real value. If you take the time to understand what the core problem really is, and define what success looks like once it’s solved, you’re much more likely to build something that matters.
The next piece is having the stomach for experimentation. Build a minimum viable product as quickly as possible, then release it, even if it’s rough. The feedback from those early users is extremely valuable. It tells you what to fix, what to double down on and what the market cares about. Your first version will almost always be your worst, and that’s okay. The key is being open to feedback, especially the negative kind.
And finally, product-market fit is still the most honest signal of market readiness. It’s not about what customers say; it’s what they do. Are they adopting the product? Are they coming back? The sooner you can get something real into their hands and observe how they use it, the faster you’ll know whether the market is ready.
Q: What do you see as the greatest challenge in aligning product development and sales teams in an AI-driven environment?
A: The biggest challenge is not having a shared understanding of the customer’s needs, the specific use case, and how success will be measured. That disconnect creates a gap between what the product team is building and what the sales team is trying to sell.
AI adds a new twist to this because it dramatically accelerates how fast you can build. AI tools can now generate entire prototypes or workflows from a prompt, so teams can spin up a minimum viable product in a matter of hours or days. But that speed can be a double-edged sword. If the initial assumptions are off, you’re just scaling the wrong solution faster.
That’s why alignment must start early. The old rule still applies: measure twice, cut once. AI can supercharge development, but if your assumptions are wrong, it’ll just get you to the wrong outcome faster.
Q: How do you ensure the next generation of marketers is ready for AI-powered product development and sales processes, beyond just education and skill development?
A: What is needed now is sharper judgment, the ability to identify what truly matters. With AI, it’s easy to get caught up in data overload, but the edge lies in zeroing in on the insights that unlock exponential value.
For instance, instead of trying to analyze dozens of customer segments, a marketer might notice that users who activate a certain feature in the first 48 hours are twice as likely to convert to paid. That single insight can guide messaging, onboarding flows and campaign strategy more effectively.
AI makes it possible to learn a vast amount about your customers, what content they engage with, how they behave across channels, what features they ignore. But this creates a risk of “analysis paralysis.” The most effective marketers will be those who can quickly separate the signal from the noise, then use AI to deepen their understanding of that signal.
A good example of this is customer feedback loops. With AI, you can process thousands of reviews or support tickets and surface common themes in minutes. Great marketers will then use that data to spot the one insight, like confusion over a pricing tier or a misunderstood value proposition, and double down on fixing that.
The future of marketing in an AI world isn’t just about being more efficient. It’s about being more focused. It’s about knowing what to ask, what to listen for, and what to act on.
Q: What do you think is the biggest misconception about AI, especially in sales and product development?
A: The biggest misconception is that AI is only for technical people. In reality, AI tools are more accessible than ever because their primary interface is now natural language. If you can describe what you want in plain English, you can use AI. This also means that more technically inclined individuals can leverage AI to quickly build solutions.
For instance, at Sprout, we initially thought that automating large portions of our payroll outsourcing operations would take a lot of time because of how complex and nuanced payroll processing can be. But our Sprout AI Labs team challenged that assumption. In just two weeks, they built bots that could automate one high-impact process. Once success was proven, the teams worked more closely to build more bots. Within three months, the bots they built automated 70 percent of manual workflows, effectively doubling client-handling capacity.
