PLATFORM · DEPLOY

Git push to production,
without the pipeline.

Scalable builds, signs, and rolls out your repo to 18 regions across 3 clouds — with health checks, region pinning, and one-click rollback. No YAML. No Terraform. No platform team.

47smedian first deploy 4smedian rollback 18regions · 3 clouds 38frameworks auto-detected
build … sign … rollout … 00:00
01Anatomy of a deploy

Every push goes through six stages.

Each stage is content-addressed, signed, and re-runnable. If anything fails, the previous version stays live — no half-rolled-out catastrophes.

Pipeline · 6 stages · 47s median

  1. 01 · 3s
    Fetch & resolve
    We shallow-clone your repo at the target ref, resolve submodules, and verify Git signatures if present.
  2. 02 · 29s
    Build the image
    We auto-detect 38 frameworks, reuse the base layer cache across your org, and build a content-addressed OCI image.
  3. 03 · 2s
    Sign with Sigstore
    Cosign + transparency log. Every image is signed, every signature is publicly verifiable.
  4. 04 · 6s
    Pre-flight checks
    Migrations, env diffs, secret rotation, and policy bundles (OPA) run before any region accepts traffic.
  5. 05 · 5s
    Rolling region-by-region
    We replace canary first, watch p95 latency and error rate for 30s, then promote across regions.
  6. 06 · 2s
    Healthcheck & promote
    Once 100% of pods report healthy, DNS swings and the old version stays warm for 10 minutes — instant rollback if anything blips.
# A real deploy log, end-to-end (47s)
$ scalable deploy --region eu,us,ap

  ● git.fetch         ✓ 3s   9c12bd4 on main
  ● build.image       ✓ 29s  bun@1.2 · cache hit 82%
  ● sigstore.sign     ✓ 2s   fulcio · rekor: 4192847
  ● preflight.diff    ✓ 6s   12 env · 1 migration
  ● rollout.canary    ✓ 5s   eu-central · p95 11ms
  ● rollout.regions
     → eu-central       ✓ 11ms  2/2 healthy
     → eu-west          ✓ 14ms  2/2 healthy
     → us-east          ✓ 18ms  2/2 healthy
     → ap-southeast     ✓ 22ms  2/2 healthy

  ✓ promoted · live in 47s
  → https://checkout-svc.scalable.app
  (rollback warm for 10m — `scalable rollback`)

$
02Built-in

The boring parts, already done.

Nine things a platform team would normally take three quarters to build. They ship in your first deploy.

⌗

Framework auto-detect

Bun, Node, Next.js, Remix, Rails, Django, FastAPI, Go, Phoenix — 38 stacks, zero config. Or bring a Dockerfile.

38frameworks supported
⌘

Base-layer cache.

We share immutable base layers across your org. Most builds finish in <30s, not because we cheat — because we don't rebuild what didn't change.

82%median cache hit rate
▢

Signed & auditable.

Sigstore signs every image. Cosign verifies in-cluster. Audit log is append-only, hash-chained, and evidence-grade.

100%of deploys signed
↗

Canary by default.

5% of traffic to the new version for 30s. Auto-rollback on p95 spike, error budget burn, or healthcheck flap.

30scanary window
↻

Rollback in one click.

Every deploy is re-runnable. Bisect a regression across 10 deploys without paging an SRE. Median 4s.

4smedian rollback
⊞

Preview env per PR.

Every pull request gets an isolated env with sampled prod data (scrubbed). Auto-destroyed 24h after merge.

1line in your CI
⊕

Migrations, guarded.

Postgres migrations run before rollout. We detect destructive operations and require an explicit flag. No surprise truncations.

0silent breaking changes
○

Secrets that rotate.

HSM-backed vault. Per-region scoping. Hot-rotate without a redeploy. Zero secrets in env files.

HSMFIPS 140-3 Level 3
⎈

BYO cloud.

Install Scalable into your AWS, GCP, or Azure account. Same two-click deploy. Your VPC, your bill.

3clouds, one control plane
03The rollback

Bisect ten deploys. Page no one.

Every deploy is content-addressed. scalable bisect binary-searches a regression across your history — and rolls back the moment it finds one.

9c12bd4feat: vector search for checkoutlive · 2m ago
3a8f10d↳ bad: p95 +180ms · error rate 2.4%18m ago
e7f4c01chore: bump bun to 1.2.41h ago
a1b9e22fix: idempotency key collision4h ago
847d12fchore: regional capacity bump9h ago
scalable bisect --good 847d12f --bad 9c12bd4 → ✓ rolled back · 4.1s

"It's the only deploy tool we trust to page no one on a Friday."

Auto-rollback isn't a checkbox — it's the default. p95 spike, error budget burn, dependency health, or a custom signal all trigger a same-second rollback. Your engineers get a notification, not a page.

↻ 4.1s median ↻ 0 false positives last 90d ↻ 10m warm window
04Region fan-out

One push. Eighteen regions.

Pick the regions per service. We pin user data to the first region listed — contractual, audited, GDPR-grade.

eu-westDUB · 6ms eu-centralFRA · 11ms us-eastIAD · 18ms us-westSJC · 24ms ap-southSIN · 22ms ap-eastSYD · 28ms sa-eastGRU · 31ms me-southBAH · 26ms
live · serving ● ring · health probe every 2s
live regions18
05Versus pipelines

What doesn't ship with Scalable.

You can keep your CI server, your platform team, and your 600-line GitHub Actions workflow. Or you can not.

GitHub Actions + Terraform
Hand-rolled
  • Pipeline length~600 lines
  • Time to first deploy3–6 weeks
  • SRE headcount2–4 FTE
  • Rollbackmanual
  • Region pinningDIY
  • Signed audit logbuild it
Render / Vercel
Hosted PaaS
  • Pipeline length0 lines
  • Time to first deploy20 min
  • SRE headcount0
  • Rollbackone click
  • Region pinningopt-in, paid
  • Signed audit logenterprise only
Scalable
Two-click production
  • Pipeline length0 lines
  • Time to first deploy47s
  • SRE headcount0
  • Rollback4s, auto
  • Region pinningcontractual
  • Signed audit logdefault
"Our deploy pipeline used to be a JIRA epic. Now it's a commit message."
DK
Daniel KovačStaff Engineer · Orbital.io · 86 services on Scalable
Deploys per week214 → 1,408
Mean time to deploy14m → 47s
Failed rollbackszero.

Your next deploy is 47 seconds away.

Start free, no card. Push a repo. Pick a region. We'll handle the rest — including the rollback, in case you need one.