Terraform plans hide what they actually change
A plan tells you the resources. It doesn't tell you which services lose quorum, which SLOs slip, or which dashboards turn red at 3am.
Now in general availability — InfraTwin 2.0
Simulate any change before you ship it.
InfraTwin is the cloud infrastructure digital twin. It mirrors your real environment and replays every Terraform, Kubernetes, IAM, and cost change against it — so you ship with evidence, not hope.
- Mirrors AWS, GCP, Azure & Kubernetes
- Read-only by design · SOC 2 Type II
- Verdict on every pull request, in seconds
Topology · payments domainLive twin · refreshed 12s ago
Healthy At risk Impacted
ALBedge.api
payments-apiEKS · prod
auth-svcEKS · prod
eventsMSK · 6 brokers
ledger-dbRDS · multi-AZ
vaultHSM-backed
warehouseRedshift
2 services impacted · ledger-db saturation +18%Blast radius: 3 services
Blast radius
3 svc+1
Detected upstream impact
Cost delta
+$1.8k1.1%
30-day forecast
62ModerateChange risk · this PR
Verdict: ship with caution
2 services impacted · 1 IAM escalation flagged
InfraTwin Review · summaryPR #1284 · feat/rate-limit-policy
This change scales payments-api to 14 replicas and upgrades the ledger DB to a 2xlarge with multi-AZ. The twin verifies p99 stays under 220ms at simulated peak. One IAM escalation introduces broader S3 delete permissions — likely unintended.
Saturation risk in fraud-scoring
Predicted CPU 81% under historical Friday peak. HPA ceiling is 18 — recommend lifting to 24 to keep p99 under 220ms.
IAM escalation on payments-ci
Role gains DeleteObject on ledger-archives/*. Not in original PR description. Recommend scoping to ledger-archives/staging/*.
Cost delta within budget
Monthly forecast +$1,820 (1.1%). Driven by db.r6g.2xlarge upgrade. Reserved-instance coverage absorbs 64%.
Suggested next steps attached as PR commentsConfidence 94%
Trusted by platform, SRE, and FinOps teams at high-stakes organizations
NBNorthwind Bank
HXHelix Health
PLPleiades Logistics
CRConvex Robotics
STStratum Telco
L&FLumen & Forge
AAAtlas Aviation
QEQuanta Energy
FCFerrous Capital
HSHyperion Studios
NBNorthwind Bank
HXHelix Health
PLPleiades Logistics
CRConvex Robotics
STStratum Telco
L&FLumen & Forge
AAAtlas Aviation
QEQuanta Energy
FCFerrous Capital
HSHyperion Studios
The cost of shipping blind
Production keeps surprising teams that shouldn't be surprised.
Cloud platforms have grown faster than the tools that review changes to them. Static plans, after-the-fact dashboards, and tribal knowledge aren't enough.
Cost shows up in the invoice, not the pull request
By the time finance flags the spike, the workload has been running for three weeks and the engineer has moved on to the next thing.
IAM drifts faster than reviews can catch
Every merge tweaks a role, a binding, a policy. Quarterly audits surface what should have been caught at PR time.
Staging never matches production
Different scale, different data shapes, different traffic. Green in staging means almost nothing about Tuesday at 2pm in production.
InfraTwin closes the gap between intent and outcome — before the deploy, not after.
How InfraTwin works
Mirror. Propose. Simulate. Decide.
Four steps. The same loop you already use in code review — applied to your entire cloud.
- 01Step 1 of 04
Mirror
InfraTwin connects to AWS, GCP, Azure, and your Kubernetes clusters. It builds a continuous, read-only model of your real environment — topology, configuration, traffic shape, cost surface, and policy state.
01
Mirror
- AWS · 3 accounts1,284 resources mirrored
- Kubernetes · 6 clusters98% topology coverage
- Datadog · live traffic shapeReplay-ready
aws.twin.connectedgcp.twin.connectedazure.twin.connected - 02Step 2 of 04
Propose
Open a pull request. Push a Terraform plan. Commit a Helm chart. InfraTwin attaches to your CI and captures the proposed change as a scenario against the live twin.
02
Propose
- PR #1284 · payments-apifeat/rate-limit-policy
- + 4 added · ~ 3 changedTerraform plan attached
- CI hook · GitHub ActionsTriggered by pull_request
aws.twin.connectedgcp.twin.connectedazure.twin.connected - 03Step 3 of 04
Simulate
We replay realistic load, dependencies, and failure modes against the twin. Reliability, latency, cost, and security signals are evaluated together — not in four separate tools.
03
Simulate
- Reliability twin · runp99 = 218ms · within SLO
- Cost twin · run+$1,820 / month
- IAM twin · run1 escalation flagged
aws.twin.connectedgcp.twin.connectedazure.twin.connected - 04Step 4 of 04
Decide
Reviewers see a single verdict: what changes, what breaks, what costs more, and which roles gain access. Approve, request changes, or block — directly from GitHub, GitLab, or Bitbucket.
04
Decide
- Verdict · ship with caution2 reviewers required
- Merge gate · blockinguntil IAM scope reduced
- Slack · platform-changesNotified · 3 owners
aws.twin.connectedgcp.twin.connectedazure.twin.connected
Terraform Twin
Terraform plans tell you what changes. We tell you what happens.
InfraTwin runs every plan against a live model of your infrastructure. Reviewers see effects on services, cost, and access — alongside the diff.
Every plan turns into a verdict
InfraTwin reads your Terraform, OpenTofu, Pulumi, or Crossplane plan and simulates it against the live twin in seconds.
Diff includes downstream effects
Not just resource changes — the services, dashboards, and policies that depend on them.
Drift detection, continuously
Catch declarations that no longer match reality before they break the next deploy.
Module-aware reviews
Reviewers see ownership and historical impact for each module, not a wall of HCL.
payments_api.tfPlan #4821 · branch feat/rate-limit-policy
4 1 3
# module.payments_api.aws_ecs_service.api
~~ desired_count = 8 -> 14
# tags propagated to task
# module.payments_api.aws_security_group_rule.ingress
-- cidr_blocks = ["10.42.0.0/16"]
++ cidr_blocks = ["10.42.0.0/16", "10.51.0.0/16"]
# module.warehouse.aws_db_instance.analytics
~~ instance_class = db.r6g.large -> db.r6g.2xlarge
~~ allocated_storage = 200 -> 600
++ multi_az = true
Twin verdict: ship with caution
Reviewed in 38sReliability Insights
Predict the incident. Then prevent it.
Latency, saturation, error budgets, and dependency blast radius — simulated together against realistic traffic shapes.
Forecast · payments-api
24-hour latency simulation, post-change
p50 p95 p99
p99 latency
218ms12ms
Error budget burn
−9%vs current
MTTR (simulated)
4m32s−42%
SLO compliance
99.93%within target
Latency forecasting
Predict p50, p95, and p99 changes per service before the change ships.
Error budget impact
See exactly how much budget a release is likely to burn under realistic load.
Dependency blast radius
Map upstream and downstream effects across every team's services.
Capacity headroom
Catch saturation risks before autoscalers fall behind real traffic.
Cost Forecasting
See the invoice before you merge it.
Every pull request gets a 30-day forecast — broken down by service, team, and commitment coverage. FinOps becomes a code-review checkbox.
30-day forecast
Monthly cost projection on every pull request, broken down by service and team.
Budget guardrails
Block merges that push a service or environment past its committed budget.
Idle and overprovisioned
Surface unused capacity continuously, not in a quarterly cleanup.
Reserved & savings plans
Forecast commitment utilization before sales pressure forces a decision.
Forecast · prod-aws
30-day cost trajectory · after this PR
Current After change
Monthly delta
+$1,8201.1%
Reserved coverage
64%+4%
Waste recovered
−$18.3kthis quarter
Idle workloads
12ready to retire
IAM & Cloud Security
Catch privilege escalations at PR time, not at audit time.
InfraTwin computes effective access before and after each change. Reviewers see who gains what — across roles, federation, and service accounts.
Effective access diff
Compare resolved permissions before and after the change, not just policy syntax.
Privilege escalation
Detect role chains that quietly grant admin-equivalent access across accounts.
Policy drift
Catch silent drift between declared IaC and the live state of your cloud.
Sensitive resource exposure
Flag any change that broadens access to data stores, secrets, or production keys.
Audit-grade trail on every approval
Every simulation, verdict, and decision is logged with attribution and timestamps.
Effective access · diffprod-aws · 5 principals affected
role/payments-cikms:eu-west-1:ledger-key
Decrypt, ReEncrypt
role/payments-cis3:ledger-archives/*
PutObject, DeleteObject
role/warehouse-runnersecretsmanager:prod/stripe-webhook
GetSecretValue
role/legacy-reportsrds:prod-ledger
Connect (legacy)
role/sre-botec2:Describe*
ReadOnly
Net change: +2 permissions · 1 removedMerge blocked pending review
Kubernetes rollout simulationeks-prod-eu1 · helm chart v4.18.2
payments-apiprod
Replicas 8 → 14 · rollout: Healthy
CPU64%
Mem71%
Saturation58%
fraud-scoringprod
Replicas 12 → 18 · rollout: Degraded
CPU81%
Mem88%
Saturation79%
ledger-writerprod
Replicas 6 → 6 · rollout: Healthy
CPU48%
Mem52%
Saturation35%
1 workload near saturation · suggest +HPA cap to 24 No PodDisruption regressions
Kubernetes Replay
Cluster changes you can trust before kubectl apply.
Schedule, scale, and rollout behavior simulated across clusters — including NetworkPolicies, service meshes, and disruption budgets.
Workload replay
Replay realistic pod scheduling, autoscaling, and disruption behavior in the twin.
Multi-cluster aware
Reason across clusters, namespaces, and federated services in one model.
Network & policy
Simulate NetworkPolicies, service meshes, and east-west traffic before rollout.
Rollout safety
Catch poison-pill manifests, drift, and broken probes before production.
AI-assisted review
An expert reviewer on every PR. Trained on your environment.
InfraTwin's assistant reads the diff, the simulation results, and your history — then writes a review the way your most senior engineer would.
Plain-English change review
Every PR gets a one-paragraph summary: what changes, what's risky, what to ask the author.
Recommended fixes
When the twin flags a regression, the assistant proposes a concrete change — not generic advice.
Root-cause hypotheses
Correlate simulated failures with the specific configuration lines that introduced them.
Faster postmortems
Replay the incident scenario in the twin to confirm what would, and wouldn't, have prevented it.
Private model · your data never leaves the region
Bring your own LLM provider or run on InfraTwin's regional inference.
InfraTwin Review · summaryPR #1284 · feat/rate-limit-policy
This change scales payments-api to 14 replicas and upgrades the ledger DB to a 2xlarge with multi-AZ. The twin verifies p99 stays under 220ms at simulated peak. One IAM escalation introduces broader S3 delete permissions — likely unintended.
Saturation risk in fraud-scoring
Predicted CPU 81% under historical Friday peak. HPA ceiling is 18 — recommend lifting to 24 to keep p99 under 220ms.
IAM escalation on payments-ci
Role gains DeleteObject on ledger-archives/*. Not in original PR description. Recommend scoping to ledger-archives/staging/*.
Cost delta within budget
Monthly forecast +$1,820 (1.1%). Driven by db.r6g.2xlarge upgrade. Reserved-instance coverage absorbs 64%.
Suggested next steps attached as PR commentsConfidence 94%
Integrations
Drops into the stack you already run.
Connect what you have. InfraTwin works with the clouds, IaC tools, CI systems, and observability platforms you already invest in.
Clouds
- AWS
- Google Cloud
- Microsoft Azure
- Oracle Cloud
IaC
- Terraform
- OpenTofu
- Pulumi
- AWS CDK
- Crossplane
Kubernetes
- EKS
- GKE
- AKS
- OpenShift
- Helm
- Argo CD
- Flux
VCS & CI
- GitHub
- GitLab
- Bitbucket
- GitHub Actions
- CircleCI
- Buildkite
Observability
- Datadog
- Grafana
- New Relic
- Honeycomb
- Splunk
Security
- Wiz
- Snyk
- Vault
- Okta
- Entra ID
- Open Policy Agent
Comms
- Slack
- Microsoft Teams
- PagerDuty
- Opsgenie
- Linear
- Jira
Data
- Snowflake
- Databricks
- BigQuery
- Confluent
- MongoDB Atlas
Need something custom? The connector SDK lets platform teams add anything with a public API.
Outcomes
Measurable change in 90 days.
What our customers report — across midmarket, enterprise, and regulated industries.
Change failure rate
−71%
Median, mid-market customers
MTTR for change-induced incidents
−54%
Across surveyed deployments
Cloud waste eliminated
23%
Average annualized recovery
Engineering hours returned
11k+
Per 100-engineer org, per year
Built for the teams who carry the pager
Different roles. Same loop. One platform.
Platform, SRE, FinOps, and security teams use the same twin — so a single simulation answers everyone's question.
Platform Engineering
Standardize how every team ships infrastructure
Give every team a paved path with simulation built in. Block unsafe patterns at PR time, not in a wiki nobody reads.
Site Reliability
See incidents before they happen
Replay realistic load and dependencies. Burn fewer error budgets. Spend on-call time on the surprises, not the avoidable.
FinOps
Stop spend at the pull request
Every change shows its monthly cost delta. Enforce budgets where decisions are made — in code review, not in a spreadsheet.
Cloud Security
Catch IAM drift before merge
Diff effective access on every change. Audit-grade evidence for every approval. Drift detection that runs continuously.
Voices from the platform
Teams who replaced anxiety with evidence.
“InfraTwin caught a Terraform change that would have removed a security group used by a service two teams away. Plan looked clean. The twin showed the blast radius in 40 seconds. That alone paid for the year.”
“Our cost reviews used to be a monthly ritual of guilt and surprise. Now the cost delta is on every PR. We've cut waste by twenty-three percent and we sleep more.”
“We replaced three tools — a drift detector, a cost forecaster, and a homegrown IAM diff bot — with one platform. Reviewers see a single verdict and the team actually trusts it.”
“We had a release that would have saturated our payments cluster on Black Friday. InfraTwin flagged it in CI. The fix was a four-line autoscaler change. Zero customer impact.”
“The IAM preview is the feature that sold the security team. We can finally see effective access change per PR, not just policy text. Audit prep dropped from two weeks to two days.”
“It feels like staging finally tells the truth. The twin gives us realistic confidence, not theater. Our change failure rate is down to under three percent.”
Enterprise & security
Architected for the environments that can't afford a misstep.
InfraTwin is built like the systems you're protecting — least-privilege, regional, auditable, and reversible by design.
SOC 2 Type IIISO 27001GDPRHIPAA-readyFedRAMP Moderate (in process)PCI DSS environment
SOC 2 Type II & ISO 27001
Independently audited controls. Annual penetration testing. Continuous monitoring of the platform that monitors yours.
Read-only by design
InfraTwin never mutates your cloud. Connector roles are least-privilege, scoped, and rotated automatically.
Regional residency
Run InfraTwin in the US, EU, or your own VPC. Configuration and telemetry never leave the region you choose.
SSO, SCIM & audit log
SAML, OIDC, SCIM provisioning, scoped API tokens, and a tamper-evident audit log. Every action attributable.
Customer-managed keys
Bring your own KMS for at-rest encryption. Disable HeapOverflow access entirely with break-glass approval.
Private deploy options
Single-tenant cloud, dedicated infrastructure, or on-prem control plane for regulated environments.
Need our security package, an architecture review, or BAA?
Trust Center has every policy, sub-processor, and audit report under NDA-free download.
Pricing
Priced per environment. Never per seat.
Reviewers, approvers, and the team running it should never be billing line items. Buy the simulation, not the chair.
Team
For growing engineering orgs adopting simulation in code review.
$2,400per environment / month
- Up to 3 cloud environments
- Unlimited users & PR simulations
- Terraform · Kubernetes · IAM · Cost
- GitHub, GitLab, Bitbucket integrations
- Standard support · email & Slack Connect
- SSO via Google Workspace, GitHub, Microsoft
Most popular
Business
For platform teams running mission-critical infrastructure at scale.
$6,800per environment / month
- Unlimited environments
- Everything in Team, plus
- Custom policy authoring & enforcement
- Advanced cost models · multi-account FinOps
- Priority support · 30-minute SLA
- SAML SSO, SCIM provisioning, audit log export
- Architectural reviews on roadmap items
Enterprise
For regulated, multi-region, or single-tenant deployments.
Customannual agreement
- Single-tenant cloud or private deployment
- Customer-managed encryption keys (BYOK)
- Regional data residency (US, EU, custom)
- Dedicated solutions engineer
- 24/7 support · 15-minute SLA · named on-call
- Procurement-friendly MSA & DPA
- FedRAMP Moderate roadmap support
All plans include unlimited users, unlimited integrations, and unlimited pull request simulations. Annual agreements available with multi-year discounting.
Frequently asked
The questions reviewers ask first.
Short, direct answers. Our docs go deeper, and our team is happy to talk specifics.
Still curious?
A solutions engineer responds within one business day.
Ready when you are
Ship the next change with evidence behind it.
Most teams open their first simulation within an hour and see measurable incident reduction in the first 90 days. We'll show you what that looks like in your environment.
- Connect your first cloud account in under an hour
- First simulation on your next pull request
- Read-only by design — your environment is untouched
No credit card. No connector that can mutate your cloud. Cancel any time.