Facets named in the 2026 Gartner® Hype Cycle™ for Platform Engineering and for Site Reliability Engineering.Facets named in the 2026 Gartner® Hype Cycle™ for Platform Engineering and for Site Reliability Engineering.
Facets
Book a Demo

AI made writing code fast.Now let it ship to production, safely and securely.

Today, every change still waits on a pipeline or a person. With Facets, AI agents ship it themselves. They can only change what you allow, and every change is checked first. Zero pipelines to manage.

Book a walkthrough

A 30-minute call. We show you how it would work with your setup and answer your security questions.

FacetsThree requests from an AI agentproduction

  • Run 5 copies of the search service instead of 3

    Your rule: scaling from 2 to 10 copies needs no approval

    Shipped on its own

  • Open port 22 to the whole internet

    Your rule: no ports open to the internet

    Not allowed

  • Delete the orders database

    Your rule: deletes need approval

    Waiting for a person

Named in the 2026 Gartner® Market Guide for AI Assistants for Infrastructure as Code

Niyo
Capillary
Purplle
Treebo
Aster
GGX
MPL
Vymo
AgNext
AskMyGuru
Astuto
CoinSwitch
CommerceIQ
Fortune 50 Automotive
FourKites
INTEGRTR
MoveInSync
PubMatic
Refold AI
Rupicard
SaaS Labs
SimpliSmart
Snabbit
Spectrum
Tekion
ZeonAI Labs
Zyla
Niyo
Capillary
Purplle
Treebo
Aster
GGX
MPL
Vymo
AgNext
AskMyGuru
Astuto
CoinSwitch
CommerceIQ
Fortune 50 Automotive
FourKites
INTEGRTR
MoveInSync
PubMatic
Refold AI
Rupicard
SaaS Labs
SimpliSmart
Snabbit
Spectrum
Tekion
ZeonAI Labs
Zyla
The problem

Every feature needs more than code.

A database, configuration, monitoring, a deployment, a route. Each has its own tool and its own pipeline. Someone has to run them in the right order and approve each step, or hand them to the ops team. So features wait, and AI-written code waits with them.

Written by AI agents24
  • Order trackingCode ready
  • Gift cardsCode ready
  • Loyalty pointsCode ready
Waiting to ship21
In your pipelines12
Checkout v21 dayStep 2 of 4: waiting for approval
With your ops team9
Search filters3 daysRSTicket open
In production3
Refund APIshipped
AI writes features in hours. They wait days in pipelines and in the ops queue.
Suyash KatyayaniCo-Founder and CTO, Purplle
“We had a great development team rolling out features but those features were not seeing the light of day because of DevOps-related problems.”
Purplle

The fix is to let AI ship too. Today, that's too risky.

Why can't your AI just ship it?

  1. 01

    It would have to run every pipeline.

    For that it needs your Terraform, the keys to everything, and the right order in every environment. One wrong step can delete a database.

  2. 02open port 22turn off encryptiondelete a databaseresize orders-db

    Guardrails only block what you thought of.

    A guardrail is a list of what AI can't do. Anything that isn't on the list still gets through.

  3. 03

    So a person checks every change.

    You're back in the queue, and it gets longer as AI writes more code.

The problem isn't your AI. It's the pipelines.

Not just infrastructure.Ship everything after the code.

Code, configuration, infrastructure, monitoring and alerts go out together in one release, with no pipelines and no hand-offs. New regions and cloud moves work the same way.

What stays
Your CIKeeps building and testing your code. Facets deploys each build.
BackstageStays your developer portal, with a Facets plugin.
Facets
Release #418Waiting
  • Code
  • Infrastructure
  • Configuration
  • Deployment
  • Routing
  • Monitoring
FirstWaitingStaging
ThenWaitingProduction
How Facets works

Build your platform from blocks your team controls.

Here's how one change moves through Facets, from your platform team to your cloud.

  1. A contract says what one piece of your platform needs, what it gives, and the rules it always keeps. AI drafts and tests it, and your team reviews it.

    Agents can use the blocks, but they can't change the rules.

    Contract · v1.4PostgresGives other services an address and passwordPlain Terraform inside, saved in your Gitpostgres/aws/1.4 · facets.yaml
    Anyone can change

    size: small or medium

    Always
    • encrypted
    • backed up daily
    • private network
    What a contract gives you
    • Compliant by defaultEncrypted and backed up, every time.
    • Mistakes caught before they runEvery choice is checked first.
    • No driftEvery change goes through Facets.
    • No Terraform, no waitingDevelopers pick what they need.
    • Connects on its ownAddresses and passwords, wired for you.
    • Write once, use everywhereOne contract for every team.

  2. In plain words, or from a menu. Either way, every option they see is one your team allows.

    Ask in plain words
    Claude Code

    >Add a medium database to checkout

    asked by a developer on the checkout team
    or
    Pick from the menu
    Add to checkout
    BlockPostgresSizesmallmediumlargenot offered by your team
    Add

  3. First it checks the request against the contract. Then it does each step, in the right order.

    Zero pipelines to manage.

    The check
    Against the Postgres contract
    • Medium is one of your options
    • It stays encrypted and backed up
    • Nothing gets deleted
    Other requests get other answers:
    • >Turn off encryptionYour contract keeps it encryptedNot allowed
    • >Delete the databaseYour rule: deletes need approvalWaits for a person
    Every step, done
    FacetsRelease #419
    1. Created the database
    2. Stored its password in your vault
    3. Connected checkout to it
    4. Released it to staging, then production
    5. Added alerts and a dashboard
    0 pipelines0 hand-offs
    plain Terraform, in your own cloud account
    Niyo
    Niyo's security team checked every change by hand for months. They never found a problem Facets had missed.
04The big unlock

Work that took a quarter now takes days or weeks.

Because every piece is a contract, a big job splits into small changes. Many AI agents can work on them at once, and Facets keeps them in order.

  1. One big job
  2. Many small changes
  3. Many AI agents at once
  4. Facets keeps the order
Integrations

Bring anything to Facets.

Your clouds, your tools and the setup you already run. Each one joins Facets, so AI can read it and act on it, and it stays the way your team set it up.

  • Clouds
    • AWS
    • Google Cloud
    • Azure
    • Private cloud
  • Infrastructure code
    • Terraform
    • OpenTofu
  • Kubernetes and deploys
    • Kubernetes
    • Helm
    • Argo CD
  • CI
    • GitHub Actions
    • Jenkins
  • Monitoring
    • Prometheus
    • Grafana
    • Datadog
  • Security and secrets
    • Vault
    • OPA
    • Falco
  • AI agents
    • Claude Code
    • Cursor
    • Praxis
    • Your own agents
  • And the restAnything with a Terraform provider can join.What you run today comes in as it is. Nothing is deleted or rebuilt.
Book a walkthroughSee a real contract in the docs

Or try Praxis, our AI agent, from your terminal:

$ brew install --cask Facets-cloud/tap/praxis

Security

Run Facets hosted by us or in your own account. Either way, your apps and data stay in your cloud.

SaaSFrom $799 a month, with a 14-day free trial

Hosted by us

Your company gets its own copy of Facets, isolated from other customers. There's nothing to install, and you connect your cloud with a script you run yourself.

Self-hostedCustom pricing

In your own cloud account

Facets runs inside your AWS, Azure or Google Cloud account, and all its data stays there. Once it's set up, you control our access: grant it for upgrades and support, remove it any time.

In production in BFSI, healthcare, SaaS and D2C.

  • BFSINiyoRupicardCoinSwitch
  • HealthcareAsterZyla
  • SaaSCapillaryTekionCommerceIQ
  • D2CPurplleSnabbit
Compare plans and pricing

  • Only your copy of Facets holds the cloud keys. AI agents never see them.

  • Every change is logged: who asked for it, a person or an agent, and who approved it.

  • Everything is standard Terraform, saved in your Git repository.

  • Single sign-on

    OktaGoogleMicrosoftand others
SOC 2
SOC 2 Type II
ISO 27001
ISO 27001 certified
GDPR
GDPR compliant
AWS Partner
AWS Partner
CNCF Silver Member
CNCF Silver Member
Google Cloud Partner
Google Cloud Partner

Teams already shipping with Facets

Read the customer stories
“We onboarded on Facets for more of a promise, but very happy to see the platform has delivered on that promise.”
Virender BishtCo-Founder and CTO, NiyoRead Niyo's story
Book a walkthrough

For the community

Questions

Book a walkthrough

You can, but the AI can still write any change your Terraform allows, and a guardrail only catches what you wrote a rule for. With Facets, AI can only make the changes your team allows, and every change is checked before it's applied.

No. Facets takes over what your Terraform already built, without recreating anything. Your own Terraform modules can become contracts.

Yes. Your CI keeps building and testing, and Facets deploys each build. Backstage stays your developer portal, with a Facets plugin. It runs on AWS, Azure, Google Cloud, OVHcloud, Vultr and Linode, and on Kubernetes clusters you run yourself.

Your CI keeps building and testing your code. After that, Facets works out every step and runs it, so there are no pipelines to write or maintain for infrastructure, configuration or deployments.

Claude Code, Cursor, Praxis (our own AI agent) or your own agents, through our command-line tool or MCP.

No. They set the contracts and decide which changes need a person in each environment.

The mistake stays inside what the contract allows. Configuration and infrastructure changes roll back. Data can't, which is why deletes wait for a person.

Yes. Start strict and move the line between what ships on its own and what waits for a person, one environment at a time.

It has for others: MPL saved 20% on cloud costs, Vymo 12%, and Purplle cut non-production infrastructure costs by 70%.

Hosted plans start at $799 a month, with a 14-day free trial. Running it in your own account has custom pricing.

Tell us what you're working on.

In 30 minutes
  1. 1You tell us what you're working on.
  2. 2We show you how Facets would work with your setup.
  3. 3We answer your security questions.

Not ready to talk? Try it free for 14 days or read the docs.

Loading