Case Study

Alpamayo logo

PREKIT — Edge Data Platform Customer Layer

End-to-end ownership of the customer-facing layer for PREKIT — Alpamayo's edge-first industrial data platform. Scope spans the Vue 3 UI, a production-grade Grafana data source plugin in Go (Phase 1 + Phase 2), the FranzAI MCP server foundation, AI assistant chat UI, Keycloak theme redesign, Sphinx documentation platform, and Storybook component library.

ClientAlpamayo GmbH
IndustryIndustrial IoT / Industry 4.0
EngagementOngoing — October 2024 to present
RoleEngineering Partner
PREKIT — Edge Data Platform Customer Layer
metric #1

Customer-facing layer owned end-to-end since Oct 2024

metric #2

Grafana plugin Phase 1 + Phase 2 shipped (Go + TimescaleDB)

metric #3

FranzAI MCP server + AI assistant chat UI live

metric #4

Sphinx docs platform adopted as standard across all Alpamayo projects

metric #5

Keycloak theme unified across full platform

01 / Brief

The Problem

PREKIT connects factory equipment — PLCs, sensors, Modbus, Siemens S7, OPC UA — into a Unified Namespace for Swiss manufacturing SMBs. As the platform grew in capability, the customer-facing layer needed to keep pace: Grafana dashboards had to serve production-scale TimescaleDB queries with access control, an MCP-based AI assistant needed a structured tool-orchestration layer, and the documentation platform had to become a reusable standard across all Alpamayo customer projects. None of these existed in a production-ready form.

02 / Constraints

Constraints

  • [01]Swiss engineering rigor — code quality, naming conventions, documentation, and review all tight
  • [02]Edge-platform context — MQTT, Modbus, OPC UA, TimescaleDB time-series at industrial scale
  • [03]Grafana plugin must work at production scale — TimescaleDB hypertable queries, dual-layer access control, audited unit tests
  • [04]AI assistant must integrate via MCP (Model Context Protocol) — structured tool orchestration, not ad-hoc API calls
  • [05]Multi-tenant authentication — Keycloak across the platform, including theme parity across products
  • [06]Documentation platform must be Dockerized, branded, and self-serve — adopted as standard across all Alpamayo projects
03 / Execution

Our Approach

Owns the customer-facing and plugin layer end to end since October 2024. Built the PREKIT Grafana data source plugin in two phases: Phase 1 MVP delivered a Go backend with direct TimescaleDB access, 4 bundled dashboards, and template variables. Phase 2 added an aggregation and downsampling UI, raw SQL editor with dual-layer access control, annotation support, credential security hardening, and full unit test coverage. The plugin is positioned for ongoing revenue-sharing as part of the PREKIT ecosystem.

Built the FranzAI MCP server foundation (OAuth 2.1, Keycloak integration) and the AI assistant chat UI with MCP-based multi-step tool orchestration — operators interact with the platform in natural language through a structured protocol layer, not raw API calls. Built a Sphinx-based Dockerized documentation platform now adopted as the standard across every Alpamayo customer project. Redesigned the Keycloak authentication theme to match PREKIT visual identity across the full platform. Maintained the Storybook component library as source of truth for the design system.

Technical Takeaway

Extending open-source visualization platforms with secure, direct database connectors and structured MCP integration allows industrial operators to query and converse with massive TimescaleDB datasets safely.

04 / Showcase

Gallery

PREKIT Hub — home & device overview

PREKIT — services configuration

PREKIT — protocol connectors (OPC UA, Modbus, S7, MQTT)

PREKIT — network topology view

PREKIT — infrastructure management

PREKIT — data model configuration

Grafana plugin — API performance dashboard

Grafana plugin — MQTT broker monitoring

Grafana plugin — host resources dashboard

05 / Impact

Outcome

  • PREKIT customer-facing layer (UI, plugins, dashboards) owned end-to-end since October 2024
  • Grafana data source plugin shipped — Phase 1 MVP (Go + TimescaleDB + 4 dashboards) + Phase 2 (raw SQL editor, dual access control, annotations, security hardening, unit tests)
  • FranzAI MCP server foundation shipped — AI assistant talks to the platform through a structured Model Context Protocol
  • AI assistant chat UI with multi-step MCP tool orchestration live
  • Sphinx documentation platform adopted as the standard across every Alpamayo project
  • Keycloak theme redesigned — authentication visually consistent across the full platform
06 / Perspective

Why this matters

Swiss industrial customers buy software as a quality signal. A Grafana plugin that handles production-scale TimescaleDB queries with auditable access control, an AI assistant routed through MCP rather than brittle glue code, and a documentation platform that every project team can rely on — these compound into credibility that matters in Swiss manufacturing SMB sales. Idin owns this layer directly, with no relay.

07 / Architecture

Tech Stack

Frontend & UI

Vue 3TypeScriptVite

Backend & APIs

GoDjangoKeycloakOAuth 2.1

Storage & Protocols

TimescaleDBMQTT

DevOps & Infrastructure

OPC UAGrafanaGrafana Plugin SDKMCP (Model Context Protocol)SphinxDockerAzure DevOpsPinia
LET'S GO

Get in touch

Have a project in mind?

Reply within 24 hours during WIB business hours. Indonesian or English — whichever fits you.

top rated plus · 100% jss · 8+ clients

PREKIT — Edge Data Platform Customer Layer | Idin Studio