design

Next-Gen UX in AI: Redefining SaaS Products in the Age of AI

Discover how AI is transforming traditional SaaS design, shifting from spreadsheet-like interfaces to adaptive, conversational, and deeply user-centric experiences. Learn about AI Assistants, Copilots, Background Agents, and more from the SaaSBhoomi talk in Chennai.

Sunit Singh

27 Mar 2025 — 6 min read

I was invited to SaasBoomi in Chennai this Tuesday to talk about the next generation of UX in AI. It was an incredible opportunity to share insights and unveil some of our ongoing work at Krutrim — work that, until that moment, had never been shown publicly.

I began by reminding the audience, “The SaaS interfaces you’re building today will soon be obsolete.” And I meant it. AI is dramatically reshaping user expectations, and we can’t keep holding on to the old ways of designing.

Why Traditional SaaS Design Falls Short

Let’s be honest: most SaaS products are basically souped-up spreadsheets. Sure, we add fancy UI components and tabular interfaces, but in the end, it’s CRUD (Create, Read, Update, Delete) under the hood. I joked on stage about how we “do our best to make it sexy,” yet we’re still bound by rigid, templated designs.

Overwhelming Choices: As I said in Chennai, “Users come with one job in mind, yet we bombard them with countless options.”
Complex Navigation: I shared my frustration with hierarchical design — digging through layer upon layer just to find one piece of information.
Manual Data Filtering: Many SaaS tools dump massive tables of data in front of us, expecting us to figure it all out ourselves.
Rigid Templates: It’s not just about convenience for developers. “Most SaaS products are template-driven, constraining true innovation and user-centric design.”

These challenges led me to emphasise the need for a new approach — one that starts, first and foremost, with the user.

Thinking User-Backwards

I firmly believe that “thinking backwards from the user never goes out of fashion — AI or no AI.” No matter how advanced technology gets, understanding the user’s life goals, KRAs, and day-to-day challenges remains critical. If we skip this step, we risk building overcomplicated features that people don’t actually need.

So before we jump to how AI can revolutionise things, we must ask:

What are users really trying to do?
What’s their environment or context?
Which steps can we streamline or eliminate?
What information do they need at their fingertips?

AI: The Great Simplifier

On stage, I played a quick audio snippet of our upcoming AI-driven news aggregator, which seamlessly stitches together trending news from multiple sources and then uses text-to-speech. The idea was to prove, in a fun way, that AI “isn’t just a chatbot or a checklist feature” — it’s a force multiplier for user experiences.

What does it simplify?

Searching → Asking: Users can simply converse with AI instead of clicking through endless drop-downs.
Static Dashboards → Contextual Insights: Instead of a page full of charts, we can serve up the right insight at the right time.
Predefined Paths → Adaptive Flows: We can guide users differently depending on context, time of day, or even the user’s mood.
Basic Personalisation → Deep Contextual Awareness: The AI can adapt its responses (and even tone) based on what it learns from the user.

Demystifying AI — The Ghost in the Machine

I often get questions about these AI “agents,” “copilots,” and “background workers.” Breaking it down:

AI Assistants: “Ask and you shall receive.”
AI Add-ons: The AI is there, but only steps in when needed.
AI Copilots: The AI actively helps you get work done — you can jump in or override it when you want.
AI Background Agents: They operate behind the scenes, synthesising data and serving insights.
AI Workers: Full-on autonomous digital workers that can replace manual tasks.

Examples shared

Evolving Interaction Models: Conversational & Adaptive UI

One of my favourite demos involved a conversational interface that not only responded in real time but also showed its “thinking” or reasoning steps. I used to say, “I often find myself speaking to an AI as though it’s human — and getting upset when it doesn’t behave!” That’s actually a good sign: it means we’re crossing the threshold into more natural, human-like interactions.

We also showcased voice-based tasks: brainstorming, dictation, and command-and-control. Yet I reminded everyone, “Voice interactions are intuitive but must balance immediacy, accuracy, and environmental constraints,” especially if your user is on the move or in a noisy office.

Adaptive UI (GenUI)

Personalisation has long been a buzzword, but too often we’ve lacked the tools to really deliver it. Now, we can feed location, time-of-day, and user history into AI systems that can dynamically shape the UI.

For instance, we took a static event-scheduling app (Sched) and let AI adapt it on the fly — knowing it was midday, the user was in Chennai, and that they might be interested in SaaSBhoomi. Instead of just showing a long schedule from 9:45 AM onward, the app automatically pinned the user’s current session to the top. “This is the time right now — here’s what’s happening,” it would say, saving them from unnecessary scrolling.

Designing for Human-AI Collaboration

“Balancing prompt-based interactions with direct manipulation is critical.” Many of us learned to trust direct control (clicking, dragging, selecting). But in an AI-driven world, we also rely on the AI to do part of the heavy lifting — and we still want the freedom to jump in and manually fine-tune something.

On stage, I demonstrated an interface with a chat-like prompt on one side and a direct manipulation canvas on the other. Sometimes you just want to click and edit something yourself; other times it’s easier to ask the AI to do it. Both approaches can (and should) live side by side.

Examples shared:

V0 — Where User & AI collaborate to build UIs with Prompts

V0 — Where User has the opportunity to co-manipulate

V0 — Where User is in full control for manipulation

ChatGPT’s Inline edit widget in the Canvas.

Final Thoughts

As I wrapped up my talk, I mentioned a small but telling story from my hotel in Chennai. I couldn’t figure out how to switch on the lights — because everything was controlled by an iPad’s digital interface. It was flashy, sure, but also confusing. “Technology without user-centric design can create more confusion than convenience.” That line drew a lot of nods from the crowd.

Ultimately, AI won’t solve every design shortcoming on its own. We must keep the user’s context, challenges, and goals at the forefront of everything we create. That’s how we future-proof our products against obsolescence. Stay curious, stay adaptive, and never forget that, at its core, UX is about helping people accomplish their goals with ease and delight.

Let’s Collaborate

If you’re a SaaS company looking to leverage AI, from leveraging compute, shifting your workloads, to using our models catalog. Krutrim would love to partner. We can help you push the boundaries of your product’s user experience. Get in touch with us at to explore possibilities.