When the agent writes half the code: rebuilding Vision Intelligence's app with Claude | Case Studies

OVERVIEW

Vision Intelligence had invested heavily in next-generation camera hardware to win larger Tier 1 and multi-facility security contracts. Its mobile app was the weak link: dated, rated 3.2 stars and unable to show the new Axis cameras alongside the Mobotix fleets customers already ran. Omnia Collective rebuilt the app in four months, bringing Claude Code into the delivery team halfway through.

Client Story

Strategy and Priorities:

Give premium hardware an app to match, so Vision Intelligence can compete for larger Tier 1 and multi-facility contracts.

Modernize the retail banking experience to retain younger demographics and increase daily app engagement.

Digital Priorities:

Rebuild the app in React Native, show Axis and Mobotix cameras side by side, and help multi-site customers manage schedule changes from one place.

Shift from a purely transactional utility to a proactive financial advisory tool.

The Challenge:

A dated app sitting at 3.2 stars, with customers adjusting security schedules one facility at a time.

Our Collaboration:

One cross-functional program covering discovery, UX research, journey mapping, a new design system and the rebuild, with Claude Code joining the team halfway through.

We partnered with the bank’s digital and data science teams to design and validate an AI-driven homepage that anticipates user needs and provides tailored financial guidance.

CONTEXT

The Challenge

Vision Intelligence's new camera hardware was built to win bigger enterprise contracts. The app customers used every day didn't match it. It was dated, sat at 3.2 stars on the App Store, and couldn't show off the new Axis cameras alongside the Mobotix fleets customers already ran. Customers running many sites were also adjusting schedules one facility at a time. The hardware said premium. The app didn't.

OUR EXPERTISE

For the first half of the program we built the conventional way. Then we set up Claude Code as an agent inside the team's GitLab pipeline. It picks up work, writes the change and opens a merge request, and an engineer decides whether it merges. More code means more to review, so the agent reviewer checks against the requirement, not just the diff, using standards written down at the organisation level and in the repository. The human approval that stays checks intent: is this still the right thing to build?

How we helped

The solution

Research and design system

Multi-source pipeline development

Discovery, UX research and journey mapping shaped a new design system built around how security teams actually use the app.

Built robust data ingestion pipelines connecting cloud platforms, CRM databases and legacy batch systems — standardising ETL/ELT processes across the organisation.

React Native rebuild

Cloud data warehouse

A full rebuild shipped to the App Store, bringing Axis and Mobotix cameras together in one experience.

Designed and deployed a scalable AWS and Snowflake architecture capable of processing billions of annual transactions — with optimised storage and compute separation for cost efficiency.

AI Scheduling Agent

Data model harmonisation

Built on Claude Opus 4.7 via Amazon Bedrock, the agent helps customers manage schedule changes across every site from one place.

Developed a high-fidelity mobile prototype featuring predictive ML spending insights, Gen-AI chat support, and dynamic UI modules.

Agentic delivery with Claude Code

Secure data sharing

Claude Code works in the GitLab pipeline and authors around half of weekly merge requests. The merge rule moved from two human approvals to one human plus the agent.

Validated the AI-driven features through extensive qualitative user testing, ensuring the algorithms felt helpful rather than intrusive.

IMPACT

The app now matches the hardware

4.3 stars

Early App Store rating, up from 3.2

50%

Of features built by Claude Code

50%

Of weekly merge requests authored by Claude Code

TOOL STACK

What powered the rebuild

Claude Code

Amazon Bedrock

React Native

GitLab

CASE STUDIES

More Success Stories

Technology Recruitment & Executive Search

Technology

57 Technology Hires at 93% Close Rate — Multi-Year Talent Partnership for a Tech Services Firm

From executive search trial to multi-year talent partnership — 57 hires delivered at 93% close rate.

Technology Recruitment & Executive Search

Technology

57 Technology Hires at 93% Close Rate — Multi-Year Talent Partnership for a Tech Services Firm

From executive search trial to multi-year talent partnership — 57 hires delivered at 93% close rate.

Technology Recruitment & Executive Search

Technology

100+ Specialist Technology Contractors Deployed Nationally for a Global Technology Leader

Five-year augmented capability partnership: 100+ specialist professionals deployed nationally to complement permanent teams.

Technology Recruitment & Executive Search

Technology

100+ Specialist Technology Contractors Deployed Nationally for a Global Technology Leader

Five-year augmented capability partnership: 100+ specialist professionals deployed nationally to complement permanent teams.

Digital Experience & CX

Healthcare

Building a Design System for Scale for a Global Healthcare and Insurance Provider

A design system built for adoption, not just documentation — giving global healthcare delivery teams the governance, contribution model and embedded support to actually use it and scale it.

Digital Experience & CX

Healthcare

Building a Design System for Scale for a Global Healthcare and Insurance Provider

A design system built for adoption, not just documentation — giving global healthcare delivery teams the governance, contribution model and embedded support to actually use it and scale it.

Back to case studies