Building Unified Customer Experience at Scale with AI: An Interview with Chirag Bansal

q01Sprinklr handles billions of social media posts daily. When someone posts about your client's brand at 3 AM, how fast does your system catch it, and what happens next?

Chirag: The intent of the entire social media infrastructure is latency. How fast can we bring messages into Sprinklr. We work with Fortune 500 companies that have set SLAs to respond to customers.

What happens is: angry customer posts a tweet → within seconds that tweet has to come into Sprinklr, be enriched by multiple AI models for sentiment analysis and tagging, go through the customer's workflow, determine if translation is needed, check if an agent is available, and decide whether it should be automated or manual engagement.

All these decisions have to be made as fast as possible, not in minutes, but in seconds. Otherwise, SLAs get heavily impacted. This is done for billions of messages across multiple channels while ensuring fallbacks if we can't fetch messages from any medium.

q02You went from Product Engineer to AVP. Do you still code, or are you now the person who decides what gets coded?

Chirag: The shift has been gradual from building to being an enabler. I still get involved in technical reviews, architecture decisions, and rolling up sleeves for critical features, but most of my energy goes into enabling my team - choosing the right problems for the right people with the right context.

The shift is from being a builder to being an enabler. I have to create an ecosystem where my team feels their productivity increases. Building is easy, honestly, but enabling is a tough task.

q03You're building AI that handles customer conversations autonomously. In 5 years, will customers even know they're talking to AI? Should companies tell them?

Chirag: If everyone is doing their job correctly, when my mom calls customer service 5 years down the line, she should feel understood and helped, it doesn't matter if it's AI or a human agent. Honestly, knowing my mom, if she knows AI can provide better answers quickly rather than waiting 20 minutes on hold, she'd prefer the bot.

But transparency is key. Trust is built when you're open about your process, not just when you solve problems. Companies should disclose when they're using AI, but it should be done naturally and not dismissively. The goal is to be natural and empathetic.

q04You led the architectural shift from monolith to microservices while handling billions of data points. How do you migrate without breaking the business?

Chirag: It's not about doing surgery on an airplane. It's about doing open heart surgery on an airplane! The key is not to treat it as rewriting the entire system. Our approach:

First, identify the most painful bottleneck and carve it out with clear boundaries. Once that's done, 50% of the battle is won. Then have proper tracking around this carved-out service. Things will break down, but the aim is to ensure they break down silently with proper rollback mechanisms and backward compatibility.

Moving from monolith to microservices isn't just a tech shift - you have to define proper SLAs, service owners. You're growing as an organization along with growing technically.

q05How are you integrating AI tools into your team's workflow? Are developers becoming more productive or just writing more code?

Chirag: For developers, AI is a tool we didn't know we needed but now can't live without. We're using AI heavily from inception (writing PRDs) to shipping features. All repetitive stuff - boilerplate code, documentation, SQL queries, test cases - everything is AI-powered to some extent.

The main thing to remember: AI gives you the first draft. You should not take its output as-is for production. We use annotations heavily, every AI-generated code is marked. This helps reviewers know they're reviewing AI-generated code, which acts as a flag.

AI is great at speeding up grunt work, but debugging edge cases and system design need strong engineering skills. The value isn't in writing more code, it's freeing up developer bandwidth to focus on core engineering problems.

q06What's your honest fear about AI tools. For example, your team becomes too dependent, or that competitors will outpace you if they use more AI tools than you?

Chirag: Both fears are correct. If my team becomes overdependent on AI without understanding technical depth, they become great engineers who can't debug what they don't understand. AI won't rescue you at 3 AM during a production outage.

On the flip side, competitors aren't just about speed, it's about innovation. The real fear isn't that AI will replace humans, but that teams who know how to use it well will ship more innovative products faster with fewer resources.

My job is to strike a balance. We enforce developers to explain AI-generated code: what was their initial thought process and why did they choose AI over their own code. This helps maintain critical thinking.

q07What about new developers, are they becoming AI prompters rather than learning fundamentals?

Chirag: New developers are definitely more curious, but on the flip side, they're learning faster than what we saw 3-4 years back. They can reverse-engineer solutions that AI suggests. If AI is used correctly, there's no alternative for critical thinking.

The key is understanding "why" more than "how." If you can reason about why a solution exists and think critically, you're doing good. But yes, some are becoming more dependent on AI for the initial thought process.

q08Where is social media technology heading? What's the next big platform shift?

Chirag: The next evolution is completely autonomous experiences. We already have AI agents interacting on behalf of brands, influencers generating AI content, and commerce seamlessly integrated into conversations.

I think the difference between support, commerce, messaging, and customer service will converge. We're moving toward a social ecosystem that's powered by AI, backed by data, and is highly customized and conversation-based. Interacting with software will be like conversation between two friends.

q09From individual contributor to AVP - What advice would you give to engineers who want to grow into technical leadership?

Chirag: The biggest shift is realizing that leadership is not about having all the answers, it's about creating an ecosystem where people can find their own answers.

Technically you have to be strong, but other skills matter: communication, trust, empathy. People never forget how you made them feel. People will go the extra mile for leaders who listen rather than just lead.

Building trust is the most important part. Stay curious, communicate clearly, look at the bigger picture, and be transparent. Success of a leader depends entirely on the success of the team.

Final Thoughts

The insights from Chirag Bansal reveal a technology landscape in rapid transformation. Social media infrastructure must scale to billions of users while maintaining millisecond latency, AI tools are transforming software development with significant productivity gains, and the future belongs to engineering leaders who can navigate complex tradeoffs between automation and human oversight.

The key takeaway? Success requires not just technical expertise but also the emotional intelligence to build empathetic, high-performing teams. As AI becomes increasingly integrated into every aspect of software development, the leaders who thrive will be those who understand that technology amplifies human capabilities rather than replacing human judgment.

Connect with Chirag: You can find Chirag on LinkedIn to continue the conversation about social media infrastructure, engineering leadership, and AI in software development.

Related Shorts

Explore short-form videos that expand on the ideas covered in this blog.