GS IT Engineering
Back to the examples

Reference project · Oracle · OCI · Agentic AI · AI-assisted development

An ops platform that makes operations visible and controllable.

Scattered checks, manual patch reviews and knowledge locked in people's heads became a central platform that shows the state of an entire server and database landscape – down to the RDS farms customers access via RDSWeb – in real time, and automates recurring operational tasks in a traceable way.

Situation

Running a grown Oracle/OCI landscape with several Windows and Linux servers and multiple Oracle databases takes time every day: checking patch levels by hand, reviewing DB health one by one, piecing together server status. On top of that come RDS farms customers use to run their applications via RDSWeb – whose availability is felt directly by the customer. The information was scattered, with no shared view.

Solution

A modular platform with clearly separated layers – each with one job:

  • Collectors

    Gather data deterministically – Oracle DB health, patch levels (Windows via WinRM, Linux via SSH/dnf) and infrastructure state.

  • Central Ops API

    FastAPI normalises every source and serves it securely (JWT + API key).

  • PostgreSQL

    The central operations database as a reliable, shared foundation for analysis and automation.

  • Dashboard

    A Vue interface with drill-downs by system, environment and severity.

  • Cost & cost centres

    Captures OCI costs and allocates them transparently to the company’s departments and cost centres – turning cloud spend into figures you can attribute.

  • Agentic AI layer

    Assesses states with LLM-assisted logic (GenAI), proposes tasks and triggers documented runbooks – such as updates within a maintenance window or server power management.

Architecture

Data flows in one direction: from the source systems through deterministic collectors into a central operations database – and from there into visualisation and automation.

A look at the dashboard

From infrastructure through the Oracle databases to the RDS farms customers use to run their applications via RDSWeb – state, load and availability at a glance. And beyond that, OCI costs cleanly allocated to cost centres.

Infrastructure & updates

24 systems monitored
3 updates pending
21 up to date
6 Oracle DBs · backup ok
Updates & maintenance
ServerTypeUpdatesStatus
app-prod-01Linux– up to date
web-prod-02Windows4 updates available
db-prod-01Linux2 security
sftp-01Linux– up to date
DB health
InstanceBackupPatch level
ORCLPROD✓ today up to date
ORCLTEST✓ today 1 patch missing
ORCLDWH✓ today up to date

RDS farms & availability

8 RDS hosts
8 reachable
142 active sessions
12 disconnected
Sessions per server
ServerFarmSessionsStatus
ts-prod-01Office38 reachable
ts-prod-02Office41 ≥ 85% RAM
ts-prod-03ERP33 reachable
ts-prod-04ERP30 reachable

OCI cost per department

€ 4,820 OCI cost (month)
100% allocated to cost centres
+6% trend vs. last month
1 department over budget
Allocation by cost centre
Cost centreDepartmentCost/monthBudget
CC-4100Production€ 1,832 within budget
CC-4200Sales€ 1,108 within budget
CC-4300IT / Operations€ 1,240 +12% budget
CC-5000Administration€ 640 within budget

Illustration with sample data – no customer reference.

Guiding principle: experience meets modern AI

„Deterministic collection, agentic assessment."

The collectors are deliberately simple and reliable; the intelligence sits in the assessment layer. This clear separation makes the platform robust, testable and extensible.

The entire platform was built with AI coding assistants – architecture, code, tests and documentation. Living proof that years of IT experience and modern AI achieve more together: experience defines the architecture and safety rules, AI accelerates the implementation.

~3 days
to the first working prototype

From the automatic data collector through the database to the dashboard with login and drill-down. It grew to today's depth of functionality in around four weeks.

Benefit

  • Operational status at a glance instead of scattered individual checks.
  • RDS farm availability in view – active and disconnected sessions, load and planned maintenance.
  • Patch and compliance transparency across Windows and Linux.
  • OCI costs made transparent – automatically allocated to departments and cost centres instead of one lump cloud bill.
  • Recurring tasks automated – with approval and audit trail.
  • A clean, versioned foundation that grows with you.

Technologies used

  • Python
  • FastAPI
  • PostgreSQL
  • Vue / Vite
  • Docker Compose
  • Oracle
  • OCI
  • WinRM / SSH

Something similar in mind for your operations? Let's talk about your server, database or automation landscape.

Arrange a consultation