Designing Sprinklr AI: Turning Widget Data into Instant, Multilingual Insights

In a dashboard ecosystem overflowing with data, users were spending too much time drilling down into widgets just to understand what the numbers meant. The experience lacked clarity, speed, and inclusivity. This case study explores how we designed Sprinklr AI — a summarization layer that transforms raw widget data into instant, multilingual insights. From uncovering pain points to crafting a popover that supports sentiment-level exploration and real-time feedback, this journey reflects how thoughtful UX can turn complexity into clarity.

Role

Role

Product Designer V

Status

Status

Live

Industry

Industry

Saas- no code/low code

Goal

Goal

Surface instant, contextual summaries directly from widgets.

Timeline

Timeline

Q2–Q3 2025

Team

Team

Product Design, PM

Context

In our dashboard ecosystem, widgets are the primary interface for users to monitor performance, sentiment, and trends across various dimensions. While powerful, these widgets often presented raw data without context, requiring users to perform manual drilldowns to interpret what the data actually meant.

This created friction, especially for users who needed quick insights or operated in multilingual environments. The challenge was to design a summarization experience that was instantintelligent, and inclusive.

The Opportunity

We envisioned a lightweight summarization layer — powered by Sprinklr AI — that could sit directly on top of supported widgets. This would allow users to access summaries with a single click, without leaving the dashboard.

90%

Adoption rate

-40%

Reduction in time to Insight

35%

Increase in engagement

Problem Statement

Time-consuming

Users had to click through multiple layers to understand widget data, slowing down decision-making.

Lack of context

Raw metrics lacked narrative, forcing users to interpret trends manually.

Limited sentiment depth

Summaries grouped mentions by sentiment but didn’t allow exploration of individual data points.

No multilingual support

Non-English users struggled with summaries that didn’t align with their platform language.

Process

To tackle these issues, we followed a structured design process.

Research & Insights

User Interviews

Revealed frustration with widget and lack of language support.

Analytics

High widget interaction, low engagement with deeper data.

Competitive Analysis

Few tools offer inline summaries or sentiment-level access.

User Quote

It takes me forever to get from the chart to the actual mentions. By the time I do, I lose the bigger picture.

User interview snapshot with Pepsi USA - Ux researcher, PM & Product Designer (Me)

Goals

Deliver instant, contextual summaries with one click.

Support multilingual summarization for global accessibility.

Provide sentiment-level exploration without losing clarity.

Reduce time-to-insight while keeping drilldown available.

The Solution

We introduced Sprinklr AI, a contextual summarization layer embedded directly into supported widgets. It’s activated via a subtle trigger icon and opens a popover that delivers intelligent, digestible insights.

Challenges & Collaboration

While we aimed to structure the widget flow for long-term scalability, we received pushback from the business.

We were working towards making space for advanced agentic capabilities. But the sprint focused towards shipping refinements for immediate rollout.

So we were holding off on foundational improvements we felt were critical for scale. Still, we optimized what we could, ensuring the experience remained clear, flexible, and ready for future evolution.

Sprint Timeframe

Sprint Timeframe

Sprint Timeframe

LLM Cost

LLM Cost

LLM Cost

IA complexity

IA complexity

IA complexity

Dev bandwidth

Dev bandwidth

Dev bandwidth

Learnings

Inline summaries work best when context-aware.

A summary without the right framing can feel shallow or misleading. Placing the AI trigger directly on relevant charts avoided overwhelming users and made insights feel naturally discoverable.

Feedback loops are essential for AI adoption

Early versions without the 👍/👎 option left users skeptical. Adding a feedback mechanism gave users control and improved confidence, while also feeding back into improving ML models.

Multilingual support is not a “nice-to-have” but a core need

Global users felt excluded when summaries defaulted to English. Enabling translations and auto-detection significantly improved accessibility and adoption in APAC and EMEA regions.

I love exploring new opportunities and meeting new people.
Let's connect!

I love exploring new opportunities and meeting new people.
Let's connect!

I love exploring new opportunities and meeting new people.
Let's connect!

Get in touch to get delightful design and ROI. Let's build it together.

Get in touch to get delightful design and ROI. Let's build it together.

Get in touch to get delightful design and ROI. Let's build it together.

© 2025 Ankur Kushwaha