DeplAI
— credits
DeplAI

Application

Dashboard

HomeYour ProfileOrganizationsUsageDocumentation

Services

UI/UX customizerSecurity AgentDASTCloudDeployInstance ManagementCode ReviewerSoonSessions

BYOK

KeysCatalogCompareUsage

Account

BillingInvoicesCreditsRefer & EarnNEWIntegrations
Settings
DE

DeplAI

Free

— creditsFree
deplaiDocumentation

Start here

Services

Account and models

Help and reference

Services

Deploy

**Deploy** turns a connected project into **AWS Terraform**, shows a **plan**, and applies **only after you confirm**. Azure and GCP may appear in planning or cost notes; **apply** in this product is **AWS only**.

Deploy turns a connected project into AWS Terraform, shows a plan, and applies only after you confirm. Azure and GCP may appear in planning or cost notes; apply in this product is AWS only.

Open Services → Deploy (/dashboard/deploy) with a project selected. After a successful apply, operate resources in Instance Management.

DeplAI generates infrastructure from repository evidence and your chosen deployment profile—not from a blank template. AWS bills your account directly; Deploy does not consume DeplAI credits for cloud usage.

---

Pipeline overview

StagePurpose
Repository AnalysisDetect languages, frameworks, databases, build/start commands
QuestionsInteractive Q&A to resolve ambiguities
Architecture DiagramVisual topology (networking, compute, data, security)
Cost EstimationMonthly estimate against your budget cap
Infrastructure GenerationTerraform bundle from profile + repo evidence
AWS ConfigRegion, credentials context, runtime inputs
App SecretsOptional application secrets for the stack
Deployterraform plan → your confirmation → terraform apply
OutputsURLs, credentials (masked), connection details

The sidebar shows your current stage. You can revisit earlier stages before apply, but confirming apply is irreversible for provisioned resources until you Destroy from Instance Management.

---

Repository analysis

Deploy inspects what is actually in the repository:

  • Application language and framework
  • Package managers and lockfiles
  • Docker/container definitions
  • Database usage and connection patterns
  • Build and start commands
  • Static vs dynamic hosting signals

Terraform is generated from this evidence combined with your deployment profile—not from a generic one-size-fits-all stack.

---

Architecture review and Infrastructure Advisor

Two paths lead to a deployment profile:

Guided architecture review

Walk through decisions for:

  • Compute — EC2 App, ECS Fargate, or Static Site patterns
  • Data — RDS, caches, object storage when detected
  • Networking — VPC, subnets, load balancers, NAT
  • Runtime — ports, health checks, environment variables
  • DNS / TLS — domain and certificate strategy when applicable

Infrastructure Advisor

Compare baseline, recommended, and resilient tiers against a monthly budget. Tiers above your cap are blocked or flagged so you cannot accidentally overspend in planning.

Choose an AWS region and whether free-tier-oriented sizing applies. Cloud (/dashboard/cloud) can supply org-approved account context when your organization has connected accounts.

---

Terraform generation

Generation combines:

  • Curated DeplAI modules (not an empty scaffold)
  • Detected databases → RDS wiring with Terraform-resolved settings
  • Profile choices (compute type, HA, scaling hints)

You receive:

  • A file bundle with main.tf, variables, outputs, and modules
  • Warnings when the repo has ambiguous or unsupported patterns
  • A run id for session tracking

Optional LLM refine

An optional LLM refine step can adjust the bundle using the same Platform / BYOK / Auto model picker as Security Agent. Invalid LLM output falls back to deterministic generators—apply never proceeds on broken Terraform.

---

Plan, confirm, apply

  1. 1DeplAI runs terraform plan with generated files, region, and AWS credentials for the run.
  2. 2Until you confirm, status stays waiting—no apply occurs.
  3. 3Confirming runs terraform apply. Progress streams in the Deploy UI.
  4. 4Failed applies remain on the Session; they are not marked successful.

Organization security gates

If your organization enables Security policy, Deploy may block confirmation until:

  • Required SAST, SCA, container scan, or DAST evidence exists
  • Critical or high findings are below configured thresholds
  • Exposed secrets are not present in recent scans
  • Production approvals count is satisfied

Fix findings in Security Agent or request an exception per your process. Details: Organizations.

---

After apply

Next stepWhere
View runtime (EC2, IPs, DNS)Instance Management → Runtime
Download Terraform / outputsInstance Management → IaC
Start / stop / restart instanceInstance Management
Tear down environmentInstance Management → Destroy
Post-deploy dynamic testDAST when verified target linked

Post-deploy security

When a verified DAST target exists, Deploy can run Post-Deploy Security after apply. Configure targets in DAST first.

---

Instance Management relationship

Deploy creates infrastructure; Instance Management operates it until destroy.

ActionEffect
StartStart stopped EC2
StopStop running EC2 (compute billing pauses)
RestartReboot instance
DestroyBest-effort deletion of DeplAI-tagged project resources (irreversible)

Details: Instance management.

---

Integrations

IntegrationRole in Deploy
GitHubSource repository for analysis
AWS credentialsPlan and apply in your account (org-approved or project-linked)
OrganizationsSecurity policy gates, cloud account approval
DASTPost-deploy security checks
BYOKOptional LLM refine during Terraform generation
SessionsHistory and logs for generate/apply runs

---

Troubleshooting

Planning and generation

SymptomWhat to check
Analysis stuck or emptyRepo access via GitHub App; try re-selecting project
No Terraform filesComplete architecture review; check warnings
LLM refine produced errorsRetry or skip refine; deterministic fallback applies
Budget blocks all tiersRaise cap or choose baseline tier

Plan and apply

SymptomWhat to check
Awaiting plan confirmation foreverRead plan output; confirm or cancel
Apply failed mid-runSessions log + AWS console for partial resources
IAM / credential errorsAWS key scope, region, org-approved account
Policy blocked applyOrganizations → Security policy; run required scans

After apply

SymptomWhat to check
Instance Management emptyApply may have failed; check Sessions
URL not reachableSecurity groups, health checks, DNS propagation
Destroy left orphansMixed AWS tagging; confirm in AWS console

Best practices

  1. 1Always read the full plan before confirming—not only the summary toast.
  2. 2Run Security Agent Full Scan before first production apply.
  3. 3Use staging + DAST Passive before production Active scans.
  4. 4Prefer re-apply from Deploy over manual console edits that drift from IaC.
  5. 5Stop instances to save cost; Destroy only when decommissioning.

---

Related documentation

  • Instance management
  • Security Agent
  • DAST
  • Organizations
  • Sessions
  • BYOK models

On this page

Pipeline overviewRepository analysisArchitecture review and Infrastructure AdvisorGuided architecture reviewInfrastructure AdvisorTerraform generationOptional LLM refinePlan, confirm, applyOrganization security gatesAfter applyPost-deploy securityInstance Management relationshipIntegrationsTroubleshootingPlanning and generationPlan and applyAfter applyBest practicesRelated documentation