Case Study - Agentic AI Pilots Integrated into Enterprise Data Landscapes

Context

The core requirement was to make agentic systems useful inside real organizations, connected to knowledge and workflows rather than operating as isolated demos.

This engagement also required tender-facing architecture work and clear scope boundaries so pilot outcomes could be assessed by technical and non-technical stakeholders.

Detailed description

A recurring delivery scenario in regulated and multi-stakeholder environments is: "prove value fast with an AI MVP, but do it in a way that can survive scrutiny and move toward production." In this case, we led the design and delivery of agentic AI MVPs, PoCs, and pilots across multiple industries, including public sector, using a full-stack approach (Python plus modern web frameworks) and stakeholder-aligned delivery practices.

The core architectural requirement was to make agentic systems useful inside real organizations, connected to knowledge and workflows rather than being a standalone demo. We architected and implemented multi-agent systems (using ADK) with retrieval-augmented generation (RAG) on Gemini Enterprise and integrated them into client data landscapes.

This engagement pattern was paired with tender-facing architecture work: we designed Google Cloud solution architectures for RfPs that resulted in deals totaling CHF 100k, demonstrating the ability to translate requirements into viable cloud architectures under procurement constraints.

On the applied GenAI side, we delivered a GenAI-powered marketing automation application that reduced content creation time from days to under one hour by leveraging Imagen, evidence of measurable cycle-time reduction rather than speculative AI value.

In parallel, we drove multiple PoCs through real user evaluation: three pharma PoCs were positively evaluated by more than 30 users and selected for pilot rollout, illustrating how we structure pilots to be testable by end users and decision-makers.

This capability is reinforced by prior delivery leadership in complex engineering organizations: leading a European frontend team in a 25-engineer organization, building an internal change-management platform, establishing frontend standards and workflows, and contributing to core platform features such as RBAC and shared UI infrastructure.

We also have hands-on evidence of shipping into production with security requirements directly with the client for Go-Live, multi-environment deployments, and measurable operational impact (reducing manual in-store inventory checks by approximately four hours per week per store via automated out-of-stock detection and alerting).

What this demonstrates for Swiss public-sector tenders: we can take an agentic AI use case from architecture to MVP, integrate it into enterprise data, validate it with users, and execute the engineering fundamentals (identity/permissions patterns like RBAC, deployment into multiple environments, and security requirements for go-live) that typically distinguish a pilot from a credible rollout path.

What we delivered

Delivery approach

We combined fast MVP execution with production-minded engineering: enterprise data integration, role and permission considerations such as RBAC patterns, multi-environment deployment readiness, and explicit security requirements for go-live.

Outcome

Three pharma PoCs were positively evaluated and selected for pilot rollout, demonstrating a repeatable path from architecture to validated pilot in high-accountability contexts.