Data Dialogues: The Pi of AI - Why Context Matters More Than Ever

Happy Pi Day! While AI and math enthusiasts are racing to uncover the latest insights, we’re leading the charge, bringing you powerful revelations on why context is the true driving force behind the future of AI.

Meet your hosts

🔹 Poonam Singh leads marketing for Catalyst, making complex tech feel simple (and actually useful). With 18+ years in B2B marketing, she’s all about cutting through the hype, helping businesses move beyond buzzwords to real-world impact.

🔹 Prathap Manohar Joshi R has been an integral part of Zoho’s AI development. As an Engineering Manager and an early contributor to AI initiatives, he has helped shape solutions like Zia Recommendation Builder. With a focus on practical applications of machine learning, he is on a mission to make machine learning not just powerful—but effortless, accessible, and game-changing for businesses everywhere!

Poonam: Joshi, Happy Pi Day! Quick test—how much of Pi can you recite before your brain gives up?

Joshi: Let’s see… 3.14159… and, uh… nope, that’s it. My brain refuses to store infinite numbers.

Poonam: Fair enough! But you know what else is infinite? Data.

You know, it's funny: When you think about how much data businesses are collecting these days; it's like Pi—it just keeps going. It's like we're all swimming in an ocean of numbers, charts, and insights. It really is. But you know, sometimes I wonder if we're actually making sense of it all.

Joshi: Exactly! It's impressive to have access to all this information, but how much of it is actually useful for making those really critical decisions that ultimately impact the bottom line? Because raw data alone doesn’t help—it’s context that makes it valuable. Otherwise, it’s just numbers floating around in the abyss.

Poonam: That’s what we’re diving into today: Contextual AI—the key to turning raw data into real business insights. Traditional AI is great at recognizing patterns, but it often lacks meaning, right?

Joshi: Yep! It can process data but doesn’t always understand intent. It's like trying to understand a word like “Jaguar” without knowing the surrounding sentence or the situation you're in. Is it the animal? The car? Or, you know, even a sports team? A basic AI model wouldn’t know. Without context, AI is just guessing.

Poonam: So, how does this lack of context actually play out in the real world? Well, I think we've all had those kind of facepalm moments where AI just completely misses the mark.

Like the other day, I got an email suggesting I buy another refrigerator. No way. I mean, I just bought one like a month ago. Or a sudden fraud alert blocking my card while I was on vacation because a purchase “looked suspicious,” and suddenly my credit card gets declined. All for a suspicious transaction that turns out to be me, you know, trying to buy concert tickets. It's so frustrating. Because the AI tool sees the unusual spending pattern. But it doesn't get that I'm on vacation in a different city.

Joshi: Oh my gosh, yes. Those can be a nightmare. I mean, who wants to go through that during a vacation. That’s why context-aware AI is the future. It doesn’t just process data—it interprets it.

Poonam: So walk me through how each of those would handle this situation. Okay. So a basic AI, you know, one that's not really considering the context. Let’s say I’m a customer, and I type into a chatbot: "This is ridiculous. I’ve been charged twice!"

Joshi: A basic AI might just pick up on those keywords. Like "charged twice." And then direct the customer to a general billing page or, you know, some frequently asked questions about payment errors. Which is not exactly what you want when you're already feeling frustrated. No, it's like adding fuel to the fire.

But how would a contextual AI approach this differently? Okay. So this is where things get interesting.

A contextual AI would actually dive deeper into the customer's interaction history. So it might see that this customer actually called customer support just a few minutes before sending that chat message. Connecting the dots, analyzing the language they used in the chat, picking up on the strong negative sentiment, the frustration. So it is understanding the emotion behind the words. And how does that translate into a better outcome? Well, because now the AI understands the urgency of the situation and the customer's emotional state. So instead of just giving them more automated responses, it can actually prioritize their issue for human escalation instead of another bot response. It can do that because it’s able to recognize that this isn't just a routine billing inquiry. This is a situation that needs immediate attention.

Poonam: Because it could escalate and potentially damage the customer relationship. And that’s the difference between keeping a customer and losing them forever.
Let's shift gears a bit and talk about something that we all do: Online shopping.
We've all been there. Browsing for a new pair of running shoes, recommendations start popping up. "Customers who bought this also bought this." "You might also like these socks." "You might also like these laces." Sometimes it's helpful if I actually need socks. But most of the time, it feels like just more digital noise.

Joshi: I agree it can be overwhelming. A basic AI will say: "People who bought this also bought that."
But imagine a world where those recommendations were actually useful.
Imagine it's a cold, rainy day outside. A context-aware AI will factor in the weather in your location, and if it’s winter, recommend you trail running shoes with better grip and water resistance even if I haven't specifically searched for those features. That's a great example of how context can make recommendations so much more relevant. It's like the AI is actually anticipating my needs. But it doesn't stop there.

There are so many other contextual clues that AI can pick up on. For example, think about your past purchases. Have you always bought Nike running shoes in the past? If so, then AI might prioritize showing you the latest Nike models or some related Nike accessories. So it's understanding your brand preferences and personalizing the experience. It could even look at what's already in your shopping cart. If you’ve already added a pair of socks, it won't keep recommending more socks. It's like having a personal shopper who actually knows your style and needs.

Poonam: That's a great way to put it. So instead of annoying customers with random recommendations, AI actually tailors suggestions to real intent, making them useful.
So we've covered customer service. We've covered ecommerce. Let's talk about another area where AI is playing a huge role: Fraud detection. Because these systems are meant to protect us. Right. How does context change security?

Joshi: Well, traditional fraud detection often relies on a set of predefined rules. A traditional fraud detection system might flag a $10,000 transaction from a new device. That might trigger an alert because it looks suspicious.
But a contextual AI would dig a little deeper and ask some more nuanced questions. Like, is this a long-standing customer with a history of large purchases? Is this new device actually their registered mobile phone? Have they recently updated their password, which might indicate they're taking security measures?
By considering all of these factors, the AI can make a more informed decision about whether the transaction is actually fraudulent. So instead of just blocking the card, it understands the full picture—reducing false positives and improving security without frustrating legitimate customers.

Poonam: That’s a much better experience for everyone involved.
So we've talked about how context can improve customer service, ecommerce, and fraud detection. Now let's talk about how businesses can use context to improve their sales and marketing efforts. Say, lead prediction because identifying those high-potential leads is critical for any business. It's all about focusing your efforts on the customers who are most likely to convert. So how can we use AI to make this process more accurate?

Joshi: Well, traditional lead prediction often relies on basic engagement metrics: email opens, downloads, or website visits. But they don't always tell the whole story.
A contextual AI would take a more holistic approach: analyze past purchase patterns—have they bought something similar before?).

Poonam: That means instead of guessing who’s a hot lead, AI actually pinpoints who’s ready to convert, which is incredibly valuable for sales teams, allowing them to focus their efforts on the most promising opportunities.

Can you talk a little bit about how QuickML makes AI more context-aware?

Joshi: So QuickML has three core capabilities that really address the need for context.

1. The first is Multi-source Data Fusion: This is all about connecting the dots. QuickML can integrate data from multiple sources—CRM logs, IoT sensors, customer support transcripts, even your marketing automation platforms. So it's bringing together all these different pieces of information to create a much more holistic view of the customer.

    

2. The second capability is Advanced ML Algorithms Suite: QuickML delivers highly personalized recommendations through a range of advanced algorithms. For example, theLightFM algorithmprovides tailored product suggestions in ecommerce by combining collaborative and content-based filtering. The SubSeq algorithm predicts the next best action for streaming platforms based on sequential user behavior. Text analytics algorithms, where text data is fed in a vectorized form to build both supervised and unsupervised NLP models.The Naive Bayes algorithm is a probabilistic classifier commonly used for text classification and spam detection.
   The Support Vector Machine (SVM) algorithm is a powerful classification technique that excels in high-dimensional spaces.
   Time series forecasting algorithms analyze historical patterns to predict future trends across various industries. Pixie, a graph-based recommendation engine, scales recommendations for social media, while the Trend Analyzer identifies emerging patterns to help industries like fashion retail stay ahead of demand. These algorithms work together to ensure contextually relevant and timely recommendations across various sectors.

    

3. The third capability is End-to-End ML Lifecycle Management–from data to deployment,Building a machine learning model isn’t just about training it once and hoping for the best. It’s a continuous process, and that’s where QuickML’s End-to-End ML Lifecycle Management comes in—it guides you through every stage, from raw data to a fully operational AI model that keeps improving over time.
   Let’s take an example.Imagine you run an online fashion store, and you want to build an AI model that predicts upcoming fashion trends based on customer purchases, social media trends, and search patterns. Here’s how QuickML helps you every step of the way:

   Data preprocessing: First, you gather and clean data from different sources—website clicks, purchase history, and even social media hashtags. QuickML helps structure all this raw data into something useful.
   Model training: Now, you train an AI model to analyze past trends and predict what styles will be hot next season.
   Model evaluation: QuickML ensures the model isn’t just making random guesses but is actually accurate. It helps you compare different models and pick the best one.
   Model deployment: Once you're happy with the model, QuickML lets you deploy it seamlessly, so it can start making predictions in real time.
   Model serving: Now, the AI is live and actively analyzing data—helping you stock up on trending products before they go viral.
   Model monitoring: But trends change fast, right? QuickML keeps an eye on your model’s performance, making sure it stays relevant. If accuracy drops, you can retrain it with fresh data.
   With QuickML, you’re not just building an ML model—you’re ensuring it keeps evolving, adapting, and delivering value over time. Whether it’s predicting fashion trends, automating fraud detection, or optimizing customer engagement, QuickML makes AI deployment effortless and scalable.

Poonam: If there’s one thing businesses should take away from this, it’s that AI is evolving beyond static rule-based automation. The future is dynamic, context-aware AI that can truly understand customers, anticipate their needs, and help businesses make smarter, faster decisions.

Joshi: Exactly! It’s the difference between a static report and a dynamic AI assistant.

Poonam: Alright, final thought: If businesses want to truly leverage AI, what’s the biggest mindset shift they need?

Joshi: I think the most important thing is to shift our mindset. We need to move away from thinking of AI as a static rule-following tool and start thinking of it as a dynamic, context-aware assistant that:
✔ Understands customer intent
✔ Predicts behavior before it happens
✔ Automates the right decisions, at the right time

Poonam: Love that. And just like Pi, which goes on infinitely, AI’s journey should be one of continuous learning and evolution.
Thanks for the great chat, Joshi!

Joshi: Always a pleasure! Now, can I finally get that Pi Day snack you promised?

Poonam: Sure—but only if you can recite 20 digits of Pi. No peeking at Google!

Joshi: … I’ll settle for coffee.

Poonam: Fair enough! I want to leave our listeners with a question to ponder. If AI can learn to understand context as deeply as we do, what new possibilities might that unlock? It's a future worth exploring.

Want to build AI that understands intent and drives real outcomes? Try QuickML.

📢 Stay tuned for Episode 3, where we’ll bust the biggest myths about AI automation—and what businesses really need to know!