AutoKaam Playbook
Coolify on Oracle ARM, the Empire Hosting Stack
The Heroku-style PaaS I actually run, on the cheapest serious cloud bare-metal in 2026.
Last reviewed:
The operator take
Coolify is the second-most important piece of empire infrastructure after the vault. It runs on a single Oracle Cloud ARM Ampere box, and behind it sit taxwallaai, papersadda, autokaam, kaam-tracker, PramaanAI, ReconTraceAI, and a dozen small services. One box. Many tenants. One UI to deploy them all.
The empire hard rule is no Vercel, no Railway, no Render, no Fly. The reason is not snobbery, it is leverage. PaaS vendors charge for the convenience of a green button, and that convenience tax across 12 empire deployments adds up to thousands of dollars a year. Coolify gives me the same green-button experience on hardware I control, and I pay only the marginal compute.
Oracle ARM Ampere free tier is the load-bearing piece. Four cores, 24 GB RAM, ARM-native. The empire runs on this for free, indefinitely, as long as the account stays active. I pay roughly Rs 1,200 a month on a paid second box for redundancy and scale headroom. Total empire hosting cost: well under a Vercel hobby seat.
The pattern: every empire app gets a Coolify Application (Nixpacks builder for most, Dockerfile for the few that need it). Push to GitHub, Coolify auto-deploys. SSL is Let's Encrypt managed by Coolify. Custom domains via Cloudflare. Backups via Coolify's S3-compatible target (I use Cloudflare R2).
The thing that took longest to learn: Coolify's environment-variable system uses an is_buildtime flag for build-time vars, and getting that flag wrong silently produces a build that runs but cannot reach its env. I lost a Saturday on this with a TaxwallaAI deploy where ANTHROPIC_API_KEY was set, but as a runtime variable, and the Next.js bundle had baked in undefined. Now I check the flag every time I add a new env.
For 2026, the Coolify alternative I half-watch is Dokploy, which has a leaner UI but a smaller feature set. I am not switching. The empire has too much muscle memory in Coolify.
The Indian-operator angle is the asymmetric advantage. Indian engineers have always been priced out of US-centric PaaS pricing; the same product that costs an SF founder pocket money costs an Indian solo founder a serious chunk of revenue. Coolify on Oracle ARM removes that wall entirely. If you are an Indian engineer and you are paying Vercel for anything past a side project, you are leaving rupees on the table.
The empire stands on Coolify. So can yours.
Why it matters in 2026
Vendor PaaS pricing crossed a threshold in 2025-26 where the convenience tax exceeds the value for serious operators. Self-hosted PaaS is now mature enough (Coolify, Dokploy, CapRover) to deliver the same green-button experience on cheap bare-metal. Oracle ARM Ampere free tier is the cheapest serious compute available, and Coolify is the cleanest way to use it.
Cost in INR
Free open source; Oracle ARM free tier covers 4-core 24GB; Rs 1,200/mo for paid scale-out instances
Use when
- +You run more than two apps and want to consolidate
- +Vendor PaaS bills are starting to bite
- +You can push code to GitHub and you want an SSL'd public URL
- +You want to control your data and runtime
Skip when
- xYou need multi-region active-active out of the box
- xYour team has zero Linux operations experience
- xYou need PaaS-grade SLA (Coolify is your responsibility)
- xCompliance binds you to specific cloud certifications you cannot get on bare-metal
Alternatives I would consider
Read next
Adjacent in the playbook
Free open source; running cost depends on host (Rs 1,200/mo Oracle ARM via Coolify is empire baseline)
FastAPI, the Empire's Default Python Serving Layer
Free open source; running cost depends on host (Rs 1,200/mo Oracle ARM via Coolify supports 5+ empire apps)
PocketBase, the Empire Backend I Run Across Every Project
Free at entry; Pro from Rs 2,000/mo (only needed for advanced WAF / analytics)
Cloudflare Pages, the Empire Static Hosting I Have Never Regretted
Free at 100K daily requests; Paid from Rs 500/mo for higher volume + larger CPU budgets