Avatus
AR & AI-powered Social Media App
Published 2024
Services
Product Architecture
Frontend Development
Backend Development
AI Integration
Custom Admin Dashboard
Timeline
Designed, developed, and launched in under 6 months
About Credence
Avatus is a cross-platform AR and AI social media app that allows users to create content using avatars instead of showing their real face. It was designed for people who want more creative control over how they present themselves online.
The product is available on iOS, Android, and desktop. It helps users generate avatar-based images, AI backgrounds, and custom edits all within the same platform.
The Problem
The Avatus team wanted to build a platform that combined:
Real-time facial tracking for AR avatars
A multi-mode camera that worked with both front and back cameras
AI-generated backgrounds based on user prompts
In-app image editing tools
Full support across iOS, Android, and desktop
Each of these functions had to work together in a way that felt smooth and easy to use. Technically, it was a complex product to build and coordinate across devices.
Our Solution
We developed the full product system that powers Avatus. That included AR face tracking, AI generation workflows, a multi-mode camera experience, and cross-platform delivery.
The app gives users three ways to create content:
Use the front camera to take a selfie with your custom mask filter, then generate a AI background using a prompt
Use the back camera to place the avatar into a scene and generate a new AI background
Create from a blank canvas by entering a prompt, then customise the image with editing tools
Each path ends with editing options like filters, contrast, brightness, and text.
What we built
1. AR Face Tracking and Avatar Overlay
Users don’t just get answers, they get custom estimates.
The AI concierge pulls real data from Credence’s systems and uses the customer’s:
Postcode
Credit score
Income and property details
to generate:
Personalised refinance or purchase quotes
Monthly payment breakdowns
Eligibility feedback based on their inputs
It’s the same logic a broker would use, but just faster and automated.
2. Multi-Mode Camera System
Credence launches individual campaigns using dedicated concierge links — each with its own messaging, tone, and targeting. The AI adapts its intro, prompts, and logic depending on where the user came from and what product they’re interested in.
Each link leads to a different version of the AI bot, depending on what the person’s interested in — like refinancing, locking in a rate, or getting a loan for a home that needs work. These links are shared through ads, emails, or social posts, so users land straight into a conversation that’s actually relevant to them.
3. AI Background Generator and Editing Tools
Credence doesn’t need a developer to update the experience. Through the admin dashboard, the Credence team can:
Upload new product info or FAQs
Train the AI with campaign-specific details
Adjust tone, prompts, and follow-up logic
Create or edit campaign links with a few clicks
Over time, the concierge gets smarter — not just from user input, but from the expanding internal knowledge base that Credence controls. All calculations and recommendations follow Credence’s internal lending criteria and compliance standards.
4. Cross-Platform Delivery
Every qualified lead — name, email, phone, credit score, quote history — is passed directly to Insellerate, Credence’s CRM.
So while the concierge handles:
Answering questions
Calculating real quotes
Collecting and validating lead info
The brokers focus entirely on high-intent users who are already halfway through the process.
What made this project complex
Synchronising AR tracking across mobile platforms
Keeping performance stable on mid-range phones
Supporting different creative modes with separate workflows
Combining real-time AR, AI generation, and editing in one interface
Making sure everything loads and responds quickly across all devices
Results and Impact
Avatus is now live and working across mobile and desktop. Users can create content without showing their face, and the platform supports full control over avatars, scenes, and final image output.
The system now powers:
Thousands of avatar selfies
AI-generated backgrounds tied to user prompts
A full set of editing tools for each image
A growing base of users exploring anonymous and creative photo sharing
Timeline
The platform was designed, developed, and launched in under 6 months. This included:
Building the AR system
Developing the multi-mode camera workflows
Integrating AI generation and editing
Launching on all platforms
Tech Specs
Frontend
SwiftUI for iOS
Kotlin for Android
React and Next.js for web
Custom camera interface supporting all three creation modes
Real-time avatar and scene previews
AR Layer
ARKit (iOS), ARCore (Android)
MediaPipe for facial landmark detection
Union Avatars for animated character overlays
AI and Editing
OpenAI for prompt-based image generation
Canvas editor with built-in tools (filters, contrast, brightness, text)
Backend Infrastructure
Node.js and Express for API services
MongoDB for storing avatars, prompts, and generated media
Cloud processing for AI tasks (AWS Lambda and GPU workers)
CDN for fast image and asset delivery across platforms
