How we leveraged multimodal AI to generate dynamic wedding invitations.
When we started building Vivah Sphere, our goal wasn’t just to make another wedding planner. We wanted to eliminate the “administrative burden” of planning. Traditionally, software requires users to navigate dozens of menus and fill out endless forms just to add a single expense or update a guest list. 🙅♂️📋
By integrating Google Gemini, we’ve shifted the interaction model from “Navigating UI” to “Natural Conversation.” 🗣️✨
1. The Death of the Form ⚰️: Agentic Chat with Gemini 2.0 Flash ⚡
For the core interaction of the app, we chose Gemini 2.0 Flash. Its high speed and low latency make it the perfect “brain” for our conversational agent. 🧠
Instead of hunting for a specific page, users simply tell the chat widget:
“I just paid ₹50,000 to the caterer for the Sangeet and added five more guests to the bride’s side.” 💬
Under the hood: ⚙️
- Intent Recognition: Gemini 2.0 Flash identifies multiple intents in a single sentence. 🎯
- Automated CRUD: The agent autonomously creates the budget entry, updates the vendor’s payment status, and appends the guest list—all without the user touching a single input field. 🪄
- Reliability: By using structured output, we ensure the AI updates our Supabase database with precision. 🛡️
2. Hybrid UX: Building for Every Generation 🤝
While we believe AI is the future, we also know that technology must be inclusive. Wedding planning often involves family members of all ages, some of whom may not be comfortable interacting with an AI agent yet. 👴👶
At Ekarna Interactive, we follow a “Dual-Path” UX Philosophy:
- The AI Path: For power users who want to move fast via natural language. 🚀
- The Traditional Path: A polished, intuitive Form UI that remains available for every feature. 📝
Every action the AI proposes is presented as an Interactive Review Card. This allows users to verify and edit details (names, dates, amounts) before confirming. Whether you use the chat or the form, the data remains synchronized, secure, and easy to manage. 🔄
3. Real-World Intelligence: Google Maps & Places Integration 🗺️
An AI is only as good as its data. To provide truly “unbiased” vendor search, we didn’t want a gated directory. We wanted the real world. 🌍
We integrated the Google Places API directly into the Gemini workflow. When a user asks for “Catering services in Perundurai,” the AI queries live Google data to provide:
- Verified business details, photos, and authentic ratings. ⭐
- Context-aware summaries of the service. 📝
- Auto-enrichment of phone numbers and addresses. 📞
4. Creative Multimodality: The “Nano Banana” Model 🎨🍌
Digital invitations are often generic. We wanted every Vivah Sphere invitation to be a work of art. 🖌️
We integrated Gemini 2.5 Flash Image (internally known as Nano Banana) into our digital invitation builder. This allows users to describe their vision—“A minimalist South Indian floral aesthetic with gold accents and marigolds”—and generate high-fidelity, photorealistic hero images instantly. 🖼️✨
Why This Matters for Your Product 🚀
Building with Gemini 2.0 and 2.5 allowed us to build faster and smarter. We didn’t have to build three different services for text, image, and search—we built one cohesive, intelligent ecosystem. 🧠🌐
At Ekarna Interactive, we don’t just build “cool tech.” We build software that respects the user’s preference, combining the efficiency of AI with the reliability of traditional design. 💎
Ready to make your SaaS AI-Ready? Contact Ekarna Interactive today → 👋