Candle Shack is a leading supplier of home fragrance and candle-making materials, trading through Shopify across the UK and EU with a seasonal, variant-rich range.
Its chief executive has set a clear ambition: to build a world class eCommerce business with AI at its core, a transformation known internally as Candle Shack 2.0. The first generation of that platform was built internally by the CEO, establishing the foundations for a wider programme of AI-enabled tools and services.
As that programme moved quickly from experimentation into production, the next priority was to strengthen the infrastructure, governance and technical leadership around it so the business could continue to develop at pace.
The Challenge
Candle Shack’s core data warehouse and ETL pipelines had been built and hosted by a specialist data business that was subsequently acquired twice. Following those changes, the infrastructure no longer had a viable long-term support route, making migration of the core PostgreSQL dataset and ETL pipelines onto infrastructure under Candle Shack’s control increasingly time-critical.
There was also an opportunity to make that data more useful operationally. Candle Shack runs two ERP instances across its UK and EU businesses. Bringing inventory information together across those systems required additional work, making it harder for the supply chain team to get a timely group-wide picture of stock and the historic trends needed to inform purchasing decisions, particularly for fragrance oils.
At the same time, Candle Shack 2.0 was developing quickly. Live data pipelines and internally developed tools were already in production, including an AI analyst that allowed staff to query data across the business. The leadership team’s view was that maintaining that pace would require more formal governance and a dedicated technical leader to take increasing ownership of day-to-day AI operations.
Vaul Labs was engaged to provide specialist support across those areas at a point when several important workstreams needed to progress quickly.
Our Approach
We worked with Candle Shack across two connected areas: data infrastructure and AI operating structure.
Data infrastructure and visibility
We first reviewed the existing data warehouse and ETL pipelines before migrating the core PostgreSQL dataset and associated pipelines from the inherited third-party environment onto Candle Shack’s own AWS infrastructure.
This removed the dependency on externally controlled infrastructure and gave Candle Shack direct ownership of an important part of its data estate.
We then built a Power BI layer over that data to bring information from both ERP instances into a clearer operational view. This gave the supply chain team improved visibility of inventory across the UK and EU businesses, alongside historic trend data to support purchasing decisions.
The work was particularly relevant to fragrance purchasing, where understanding inventory positions and demand trends across a large and seasonal range is important to making timely buying decisions.
The team was also provided with the documentation and knowledge needed to operate the new environment.
AI governance and leadership
The second area of work focused on putting more structure around a fast-moving AI programme.
We ran structured discovery across ten departments, documenting 49 cross-functional processes and identifying four recurring themes. This work informed an AI operating model covering two complementary areas.
Build governance: a five-stage production process with AI-specific checkpoints, a build inventory and a production-readiness framework covering areas such as ownership, security, code quality and data handling.
User governance: an acceptable-use framework, tool approval process, AI literacy programme and guidance relating to the company’s responsibilities under the EU AI Act.
The work gave the executive team a common framework for discussing how AI should be developed, governed and moved into production as the programme expanded.
We also supported Candle Shack in recruiting its first Head of AI Operations. This included helping shape the role around the company’s requirements, carrying out a technical interview and providing the leadership team with an independent assessment of technical fit.
Outcomes
“Vaul Labs supported us at a point when Candle Shack 2.0 was moving very quickly and we wanted to put the right governance and technical leadership around that momentum. They helped us bring our core data pipelines onto infrastructure we control, gave our supply chain team better visibility across two ERP systems, and supported us in putting more structure around AI governance and recruiting our first Head of AI Operations. They were responsive when we needed to move quickly, and their support helped us progress several important foundations while the wider transformation continued at pace.”
Duncan MacLean, Chief Executive, Candle Shack
From external infrastructure dependency to direct ownership. Candle Shack’s core data warehouse and ETL pipelines now operate within infrastructure controlled by the business, removing an increasingly time-sensitive dependency on a third-party environment.
Improved visibility across two ERP systems. The Power BI layer brought inventory data from both businesses into a more accessible cross-company view and added historic trend information to support purchasing decisions, particularly within fragrance buying.
More structure around a rapidly developing AI programme. Candle Shack established an initial governance framework covering both the development of AI systems and their wider use across the organisation, giving the executive team a clearer structure for oversight as the programme grows.
Dedicated leadership for the next stage. Candle Shack appointed its first Head of AI Operations, with Vaul Labs supporting the definition and technical assessment of the role. The business is now building further internal capability around that position.
Across the engagement, Vaul Labs provided specialist support at a point when Candle Shack was moving quickly and several important foundations needed to be addressed in parallel.