Insights
Engineering notes from our work with Claude, enterprise AI, and regulated software.
April 25, 2026·6 min read
Turning Claude Code into an engineering platform: our experience with claude-enterprise-templates
Handing a team access to Claude Code is not enough. Without structure it turns into chaos. How we approached the problem as engineers — through standardisation, roles, and architecture.
Claude CodeCLAUDE.mdSkillsEnterpriseAutonomous AI
March 3, 2026·8 min read
MCP + Greenplum in Banking: how to give LLM access to DWH without breaking Data Governance
We are replacing a 14-year-old Oracle DWH with an MPP platform on Greenplum / Arenadata DB under full Data Governance. Model Context Protocol turns LLM access to the warehouse from an ungoverned shortcut into a contract-based API surface.
MCPGreenplumData GovernanceBanking DWH
February 14, 2026·7 min read
Anonymizing PII before it reaches an LLM: lessons from a Kazakhstan financial holding
A reversible proxy that anonymizes incoming messages, calls the upstream model, and restores values for the user. What worked, what we underestimated, and when this kind of project is infrastructure rather than 'AI practice'.
LLM ProxyPIIKZ_IINFastAPIPresidio