DISCOVER PROJECT

DISCOVER PROJECT

SRIPTO

KANVA

Kanva AI-Powered Virtual Human Assistant System
Kanva AI-Powered Virtual Human Assistant System

2026

2026

2026

SERVICES

Product Design, System Architecture, UI/UX Design, Visual Interaction Design, 3D Avatar Experience

— Human–AI interaction is entering a new era. Kanva, powered by SRIPTO Avatar Runtime (SAR), is a hyper-realistic virtual human assistant that communicates through synchronized 3D facial expressions, gestures, and natural language. As the lead designer and product architect at SRIPTO Technologies, I transformed Kanva from an ambitious concept into a performant, emotionally aware, and visually immersive product that delivers humanlike conversational experiences across devices.
— Human–AI interaction is entering a new era. Kanva, powered by SRIPTO Avatar Runtime (SAR), is a hyper-realistic virtual human assistant that communicates through synchronized 3D facial expressions, gestures, and natural language. As the lead designer and product architect at SRIPTO Technologies, I transformed Kanva from an ambitious concept into a performant, emotionally aware, and visually immersive product that delivers humanlike conversational experiences across devices.
— Human–AI interaction is entering a new era. Kanva, powered by SRIPTO Avatar Runtime (SAR), is a hyper-realistic virtual human assistant that communicates through synchronized 3D facial expressions, gestures, and natural language. As the lead designer and product architect at SRIPTO Technologies, I transformed Kanva from an ambitious concept into a performant, emotionally aware, and visually immersive product that delivers humanlike conversational experiences across devices.

Project Overview

The Kanva project demanded uncompromising fidelity in emotional expression, latency, and conversational intelligence to satisfy SRIPTO’s technical and UX goals. The objective of Kanva was to create a virtual assistant that feels present — visually expressive, contextually aware, and emotionally adaptive — while remaining lightweight enough for real-world deployment. The goal was to design a system that blends cinematic 3D rendering with advanced conversational AI, delivered through SRIPTO’s SAR runtime for synchronized avatar rendering and speech. Collaborating with SRIPTO’s AI and graphics teams, I led the design efforts to ensure Kanva provided a compelling, natural, and trustworthy user experience.

The Kanva project demanded uncompromising fidelity in emotional expression, latency, and conversational intelligence to satisfy SRIPTO’s technical and UX goals. The objective of Kanva was to create a virtual assistant that feels present — visually expressive, contextually aware, and emotionally adaptive — while remaining lightweight enough for real-world deployment. The goal was to design a system that blends cinematic 3D rendering with advanced conversational AI, delivered through SRIPTO’s SAR runtime for synchronized avatar rendering and speech. Collaborating with SRIPTO’s AI and graphics teams, I led the design efforts to ensure Kanva provided a compelling, natural, and trustworthy user experience.

The Kanva project demanded uncompromising fidelity in emotional expression, latency, and conversational intelligence to satisfy SRIPTO’s technical and UX goals. The objective of Kanva was to create a virtual assistant that feels present — visually expressive, contextually aware, and emotionally adaptive — while remaining lightweight enough for real-world deployment. The goal was to design a system that blends cinematic 3D rendering with advanced conversational AI, delivered through SRIPTO’s SAR runtime for synchronized avatar rendering and speech. Collaborating with SRIPTO’s AI and graphics teams, I led the design efforts to ensure Kanva provided a compelling, natural, and trustworthy user experience.

Working closely with engineering, we integrated Kanva with BRAVO (SRIPTO’s chatbot core) and SAR for end-to-end orchestration — from intent parsing to real-time facial rendering and lip sync. The final product offers seamless conversational flow with synchronized visual responses that enhance trust and engagement.

My responsibilities included defining the overall product UX, designing avatar interaction patterns, prototyping motion and expression systems, and validating the experience through iterative user testing and performance tuning. We focused on a cohesive design language that made complex AI behaviour feel familiar and approachable.

Working closely with engineering, we integrated Kanva with BRAVO (SRIPTO’s chatbot core) and SAR for end-to-end orchestration — from intent parsing to real-time facial rendering and lip sync. The final product offers seamless conversational flow with synchronized visual responses that enhance trust and engagement.

Working closely with engineering, we integrated Kanva with BRAVO (SRIPTO’s chatbot core) and SAR for end-to-end orchestration — from intent parsing to real-time facial rendering and lip sync. The final product offers seamless conversational flow with synchronized visual responses that enhance trust and engagement.

My responsibilities included defining the overall product UX, designing avatar interaction patterns, prototyping motion and expression systems, and validating the experience through iterative user testing and performance tuning. We focused on a cohesive design language that made complex AI behaviour feel familiar and approachable.
My responsibilities included defining the overall product UX, designing avatar interaction patterns, prototyping motion and expression systems, and validating the experience through iterative user testing and performance tuning. We focused on a cohesive design language that made complex AI behaviour feel familiar and approachable.

Team refining emotional interaction models

Process

The design process balanced creative exploration with strict performance constraints dictated by real-time rendering and low latency expectations. We started by building detailed interaction personas and conversation scenarios to cover casual chat, technical help, mental wellness check-ins, and transactional flows. These scenarios guided the creation of animation libraries and expression taxonomies that SAR could blend in real time.

The design process balanced creative exploration with strict performance constraints dictated by real-time rendering and low latency expectations. We started by building detailed interaction personas and conversation scenarios to cover casual chat, technical help, mental wellness check-ins, and transactional flows. These scenarios guided the creation of animation libraries and expression taxonomies that SAR could blend in real time.

High-fidelity prototypes were created to validate lip sync, gaze behavior, and micro-gestures. Parallel engineering sprints optimized the speech-to-animation pipeline and network streaming to keep response times under strict thresholds. Regular user testing sessions measured perceived empathy, clarity, and comfort, and insights from those tests were fed back into iterated animation timing and dialogue pacing. Throughout the project, collaboration between product, animation, ML, and backend teams was constant. The iterative cadence ensured Kanva remained emotionally expressive while meeting engineering budgets for memory, CPU/GPU use, and bandwidth. The final process delivered a product that balanced beauty, believability, and reliability.

High-fidelity prototypes were created to validate lip sync, gaze behavior, and micro-gestures. Parallel engineering sprints optimized the speech-to-animation pipeline and network streaming to keep response times under strict thresholds. Regular user testing sessions measured perceived empathy, clarity, and comfort, and insights from those tests were fed back into iterated animation timing and dialogue pacing. Throughout the project, collaboration between product, animation, ML, and backend teams was constant. The iterative cadence ensured Kanva remained emotionally expressive while meeting engineering budgets for memory, CPU/GPU use, and bandwidth. The final process delivered a product that balanced beauty, believability, and reliability.

High-fidelity prototypes were created to validate lip sync, gaze behavior, and micro-gestures. Parallel engineering sprints optimized the speech-to-animation pipeline and network streaming to keep response times under strict thresholds. Regular user testing sessions measured perceived empathy, clarity, and comfort, and insights from those tests were fed back into iterated animation timing and dialogue pacing. Throughout the project, collaboration between product, animation, ML, and backend teams was constant. The iterative cadence ensured Kanva remained emotionally expressive while meeting engineering budgets for memory, CPU/GPU use, and bandwidth. The final process delivered a product that balanced beauty, believability, and reliability.

Kanva launched as a versatile virtual assistant that provides humanlike, emotionally adaptive interactions across platforms. Every design decision focused on making conversations feel natural, safe, and useful.

Final Design

Hyper-Realistic 3D Avatar Rendering (via SAR)

Hyper-Realistic 3D Avatar Rendering (via SAR)

Hyper-Realistic 3D Avatar Rendering (via SAR)

Emotion-Adaptive Response Engine

Emotion-Adaptive Response Engine

Emotion-Adaptive Response Engine

Real-Time Lip Sync & Facial Rendering

Real-Time Lip Sync & Facial Rendering

Real-Time Lip Sync & Facial Rendering

Cross-Platform Performance & Graceful Degradation

Cross-Platform Performance & Graceful Degradation

Cross-Platform Performance & Graceful Degradation

Visual continuity, refined motion curves, and contextual expression rules ensure that Kanva’s reactions are coherent and meaningful. Accessibility and fallback modes were included so that devices with limited resources receive an equivalent conversational experience (audio + simplified avatar or audio-only), preserving usability without sacrificing personality.

Visual continuity, refined motion curves, and contextual expression rules ensure that Kanva’s reactions are coherent and meaningful. Accessibility and fallback modes were included so that devices with limited resources receive an equivalent conversational experience (audio + simplified avatar or audio-only), preserving usability without sacrificing personality.

Visual continuity, refined motion curves, and contextual expression rules ensure that Kanva’s reactions are coherent and meaningful. Accessibility and fallback modes were included so that devices with limited resources receive an equivalent conversational experience (audio + simplified avatar or audio-only), preserving usability without sacrificing personality.

The final design for Kanva harmonized cinematic visuals with pragmatic UX. SAR powers a responsive avatar rendering layer that synchronizes animations to BRAVO’s generated responses, producing believable facial motion, gaze, and gestures with minimal latency. The interface layers expose simple controls for adjusting Kanva’s expressiveness and privacy modes, making the experience comfortable for diverse user preferences.
The final design for Kanva harmonized cinematic visuals with pragmatic UX. SAR powers a responsive avatar rendering layer that synchronizes animations to BRAVO’s generated responses, producing believable facial motion, gaze, and gestures with minimal latency. The interface layers expose simple controls for adjusting Kanva’s expressiveness and privacy modes, making the experience comfortable for diverse user preferences.
The final design for Kanva harmonized cinematic visuals with pragmatic UX. SAR powers a responsive avatar rendering layer that synchronizes animations to BRAVO’s generated responses, producing believable facial motion, gaze, and gestures with minimal latency. The interface layers expose simple controls for adjusting Kanva’s expressiveness and privacy modes, making the experience comfortable for diverse user preferences.

LET'S MAKE SOME DIFFERENCE
BY BEING DIFFERENT.

- Sri Vardhan Yeluri

LET'S MAKE SOME DIFFERENCE
BY BEING DIFFERENT.

- Sri Vardhan Yeluri

LET'S MAKE SOME DIFFERENCE
BY BEING DIFFERENT.

- Sri Vardhan Yeluri

Create a free website with Framer, the website builder loved by startups, designers and agencies.