← All case studies
Product engineering collaboration · Arctellis

From concept to launch in eight weeks

Arctellis is an AI-powered commercial intelligence platform. Its initial working version was developed and launched within eight weeks. Barrington contributed to the architecture and engineering of its production intelligence platform across architecture, data engineering, AI product design and application build. It follows the platform from the build to the business value it now delivers for a UK spirits business.

Contribution
Architecture and engineering
Status
Live in production

Barrington has contributed architectural and engineering expertise.

  • 8 weeks

    From concept to initial launch

  • 2–3 days → ~2 min

    Monthly management reporting

  • 3 systems

    Operational source systems integrated

01 · Challenge

Information spread across too many systems

A UK spirits business ran its management reporting by hand. Its information sat across point-of-sale, booking, production-management and finance tools, and spreadsheets; a management view meant extracting, reconciling and re-keying data every month — work that consumed two to three days.

02 · Our contribution

Building the platform

Arctellis keeps the systems a business already uses and adds a governed intelligence layer across them: one commercial model, agreed definitions, and several ways to use the same answers.

  • Secure, cloud-based workspace: a multi-tenant Azure platform with isolated client workspaces, Microsoft 365 sign-in and backend-enforced tenant boundaries. The web application, API and data pipeline are deployed separately and communicate through a queue.
  • Data integrations: connectors and managed uploads bring point-of-sale, booking and production-management data into a canonical commercial model, keeping lineage back to the source row.
  • Dashboards: governed commercial KPIs, year-on-year comparison and breakdowns by channel, division, product and customer, exportable to Excel.
  • Automated reporting: monthly management packs are generated in the background from the governed model, validated figure by figure, rendered as Excel workbooks and kept with a full audit record.
  • Conversational data access: plain-English questions are answered through approved query tools. The model never writes SQL, and it says when data for a period is not loaded rather than reporting a misleading zero.
04 · Outcome

Delivering business value

The customer confirms that monthly reporting which previously took two to three days now takes about two minutes, generated on demand from the governed model.

Management information is now available when it is needed, rather than when someone has had time to prepare it, so the team spends less time assembling information and more time acting on it.

Dashboards, conversational queries and reports draw on the same governed definitions, so the figures agree wherever they appear, and management can ask cross-cutting questions — which channels are growing, which customers matter most — without a manual hunt.

Platform
Arctellis
Relationship
Product engineering collaboration
Outcome
Management reporting · AI applications · Forecasting and commercial decisions
Start here

Facing something similar? Find a practical starting point.

Book an advisory session

A dedicated hour with an experienced advisor.