At a glance

  • A governed context platform gives Claude access to Bora Brasil’s business data and corporate documents, with every answer traceable to its source.
  • 83 AI opportunities mapped across the business, narrowed in three rounds to 5 priority use cases for the first-year roadmap.
  • 62 employees live on Claude, scaling to up to 150, with licences allocated by role, area and seniority.
  • Around 60 employees trained hands-on, on live company data, in the first cohort.

The challenge

Bora Brasil was formed from the merger of two family-owned haircare brands, Skala Brasil and Lola From Rio. AI adoption had started bottom-up: around 15 early users were already working with Claude, connected to their email, calendar, Teams and files. Leadership wanted to turn that momentum into a company-wide capability. That meant answers grounded in the group’s own data, access that respects existing permissions, and a governed environment so that sensitive commercial information does not drift into unsanctioned AI tools.

The approach

Over 9 weeks, Artefact and Bora Brasil worked across three streams, combining fast bottom-up adoption with a top-down roadmap and governance.

1. A context platform on Bora Brasil’s own data

Artefact designed and built a context platform on Azure, approved by Bora Brasil’s technical leadership. Claude reaches the group’s data through a single MCP server:

  • Structured data in Microsoft Fabric: sales figures, KPIs and budget data.
  • Unstructured documents in SharePoint: contracts, policies, presentations and commercial documentation.

A business knowledge layer holds the official glossary, KPI definitions and business rules. Each document is indexed with a semantic description, so Claude can find the right file and read it in full. Access always mirrors each employee’s existing permissions, and every answer cites the table or document it came from.

2. Prioritized use cases

Through interviews with directors across Commercial, Marketing, Innovation and R&D, Operations, Finance and HR, the team mapped 83 AI opportunities. A first strategic filter on feasibility and expected value brought this down to 33. Detailed sizing with the directors then produced 24 use cases across 8 business areas, each scored on business value (rated by the business) and technical complexity (rated by the technical team).

A make-or-buy assessment identified 15 cases to build and 9 that existing systems or market solutions already cover, so investment goes only where Bora Brasil’s own data and processes make the difference. A final prioritization workshop with the leadership team selected the 5 use cases that make up the year-one roadmap.

3. An AI Office to run it

The operating model is a central AI Office with light spokes in each business area, represented by AI Champions. The AI Office covers four areas:

  1. Leadership and governance: owns the AI portfolio and licence allocation.
  2. Use cases: a squad that builds the prioritized use cases.
  3. Platform: a team that maintains the context platform, security and core AI tools.
  4. Adoption: change management and training, in partnership with HR.

A usage policy and best-practice guide keep AI work inside the approved Claude environment.

4. Hands-on adoption on real data

Training ran entirely on Bora Brasil’s live data rather than generic demos. In a 150-minute session, each participant answered a real business question: where Skala Brasil is winning and losing sales year over year, and where the sales team should focus next. Along the way, participants:

  • set up a Claude Project with fixed business rules;
  • pulled official KPI definitions from SharePoint;
  • had Claude plan its queries before running them against Fabric;
  • built an interactive sales dashboard as a shareable Claude artifact;
  • validated where every number came from before sharing it.

Faster participants turned the dashboard into a weekly scheduled routine in Claude Cowork. The training itself was delivered through an interactive guide built with Claude, with a live facilitator view tracking each participant’s connection status and progress.

Results

  • 62 employees live on Claude, scaling to up to 150
  • Around 60 employees trained in the first cohort on 6 October, with a second cohort on 13 October
  • Each participant built a working sales dashboard on live company data in a single 150-minute session
  • 83 → 33 → 24 → 5: a business-validated AI roadmap delivered in 9 weeks

Early value, measured by the early adopters:

  • In a first survey of 30 licence holders, employees reported saving at least 4 hours per week on average, about 10% of a standard working week. Seven in ten save at least three hours, and one in six save more than ten.
  • Across the group, that adds up to a conservative floor of roughly 120 hours returned to the business every week, or more than 6,000 hours a year. Theses hours can now be dedicated to more strategic assignments.
  • The use cases behind these numbers are concrete:
    > In HR, a program to convert payroll data-load layouts, which an external consultancy had sized at around 40 hours, was built in 90 minutes.
    > An automated timekeeping report cut about 8 hours of manual checking and gave managers a faster view of overtime, absences and delays.
    Rebuilding the absenteeism indicator saved another 6 hours, and the personnel-package analysis that used to take two days is now done in a few hours.
    > In Finance, a Serasa credit check embedded in Pipefy removes manual client lookups, and a scheduled sales report now reaches managers without anyone building it by hand.

What’s next:

The AI Office is now scaling usage across teams. Next steps are building company-specific Claude Skills on top of the context platform and moving the 5 prioritized use cases into delivery.