AI-assisted, local-first, multi-cloud waste detection — that never touches your infrastructure.
Find idle VMs, orphaned disks, oversized resources and other cloud waste across your accounts, track findings over time, and get an AI-written executive summary — all from a tool that runs entirely on your own machine.
🇹🇷 Türkçe bölüm aşağıdadır. / Turkish section below.
Most cloud cost tools want an agent in your environment, a SaaS account, and write access to your subscriptions. That is a hard "no" for many security-conscious teams — and an unnecessary risk for a job that is fundamentally read and report.
Bigs Cloud Optimizer takes the opposite approach:
| Principle | What it means in practice |
|---|---|
| 🔒 Local-first | Runs on your laptop/workstation. Credentials never leave your machine. No SaaS, no data upload. |
| 👁️ Read-only by design | Every policy is scan-only — there are no actions: blocks anywhere. Nothing in your cloud is ever modified. |
| 🧭 Human-governed | Findings are recommendations, not automations. Remediation decisions stay with people, scored against the Bigs RCI framework (Predictability, Reversibility, Blast Radius). |
az login (your session)
│
▼
┌─────────────────┐ ┌──────────────────────┐ ┌─────────────────────┐
│ Cloud Custodian │───▶│ SQLite Operational │───▶│ HTML report │
│ read-only │ │ Memory (local file) │ │ + before/after drift │
│ policies/*.yml │ │ every scan appended │ │ + AI exec summary* │
└─────────────────┘ └──────────────────────┘ └─────────────────────┘
* optional — works with Anthropic, OpenAI or Gemini keys; sends only the metadata already visible in the report, never credentials or raw resource JSON.
The part most tools miss: every scan is stored in a local Operational Memory. Run the same scan next week and the report automatically marks findings as new / resolved / persisting — so you see whether waste is actually being cleaned up, not just a snapshot.
New machine? See SETUP.md for Windows / Ubuntu prerequisites.
az login
python scripts/run_scans.py --label "Customer Name"
python scripts/generate_report.pyOpen the generated reports/report-<timestamp>.html in a browser. Re-run on
your next visit to build the before/after history.
More options (single policy, account selection, service principal, AI summary)
Run a single policy:
python scripts/run_scans.py policies/azure-idle-vms.ymlWhich Azure account is scanned: by default, the active Azure CLI session
(az account show). To scan another subscription:
az account set --subscription <subscription-id>If a .env file exists (copy .env.example) with AZURE_TENANT_ID /
AZURE_CLIENT_ID / AZURE_CLIENT_SECRET / AZURE_SUBSCRIPTION_ID — a
read-only (Reader) service principal — it takes precedence over the CLI
session. Pass --cli-auth to force the CLI session for one run. Neither is
required.
AI executive summary — set in .env:
AI_PROVIDER=anthropic # anthropic | openai | gemini
ANTHROPIC_API_KEY=sk-ant-...
- Business-facing overview: readme.txt
- Architecture, decisions, roadmap: bigs-cloud-optimizer-brief.md
- Machine setup: SETUP.md
Built by Bigs Bilişim — an Istanbul-based engineering company applying 26+ years of enterprise systems experience to AI-assisted IT operations. This tool shares its governance DNA with our AI-driven monitoring operations platform: AI proposes, deterministic rules constrain, humans approve, records hold everyone accountable.
Want a walkthrough on your own subscription? Reach out via bigsbilisim.com or LinkedIn — a first scan takes about 15 minutes.
Yapay zekâ destekli, tamamen yerel çalışan, çok bulutlu israf tespiti — altyapınıza asla dokunmadan.
Atıl VM'leri, sahipsiz diskleri, gereğinden büyük kaynakları hesaplarınız genelinde bulur, bulguları zaman içinde takip eder ve yapay zekâ yazımı bir yönetici özeti üretir — hepsi kendi makinenizde çalışan bir araçla.
Çoğu bulut maliyet aracı ortamınıza ajan, SaaS hesabı ve yazma yetkisi ister. Bigs Cloud Optimizer tersini yapar:
- 🔒 Local-first: Kimlik bilgileri makinenizden asla çıkmaz. SaaS yok, veri yükleme yok.
- 👁️ Tasarım gereği salt-okunur: Tüm politikalar yalnız tarar — hiçbir yerde
actions:bloğu yoktur, bulutunuzda hiçbir şey değiştirilmez. - 🧭 İnsan yönetiminde: Bulgular öneri niteliğindedir; müdahale kararı insanda kalır ve Bigs RCI çerçevesiyle (Öngörülebilirlik, Geri Alınabilirlik, Etki Alanı) puanlanır.
Yeni makine kuruyorsanız önce SETUP.md (Windows / Ubuntu ön koşulları).
az login
python scripts/run_scans.py --label "Musteri Adi"
python scripts/generate_report.pyÜretilen reports/report-<zaman>.html dosyasını tarayıcıda açın. Bir sonraki
ziyarette aynı komutları tekrar çalıştırın — rapor bulguları otomatik olarak
yeni / çözülmüş / devam eden olarak işaretler; israfın gerçekten
temizlenip temizlenmediğini görürsünüz.
Ayrıntılar (tek politika çalıştırma, hesap seçimi, servis prensibi, yapay zekâ özeti) için yukarıdaki İngilizce "More options" bölümüne bakın — komutlar birebir aynıdır.
Bigs Bilişim — İstanbul merkezli, 26+ yıllık kurumsal sistem deneyimini yapay zekâ destekli BT operasyonlarına uygulayan mühendislik şirketi. Kendi aboneliğinizde 15 dakikalık bir tanıtım ister misiniz? bigsbilisim.com veya LinkedIn üzerinden ulaşın.