We deploy and operate QM — Y Combinator's open-source agent harness for startups. Every employee gets an isolated workspace. Teams collaborate in Slack and on the web. Infrastructure stays in your VPC, on your bill.
Plug in your team size and engineer rate in our cost calculator to see first-year DIY vs managed totals — including AWS and your team's time.
Wondering what DIY actually costs in year one? Run the cost calculator — model engineer hours, AWS spend, and the pilot offer for your headcount.
QM's AWS target uses Terraform, ECR, RDS, S3, and Lambda MicroVMs for per-scope agent sandboxes — the same path YC documents for production deployments.
Sessions, memory, queue, and org state — durable and backup-ready in your account.
Sandbox home snapshots, blobs, and artifacts — encrypted, lifecycle-managed.
Isolated agent "computers" per person and Slack room — persistent disk, strong isolation.
Immutable control-plane and sandbox images pinned by digest — reproducible deploys.
Infrastructure as code in a deployment repo you own — plan, apply, and audit every change.
Digitize operates via a scoped IAM role with external ID — no shared root credentials.
System topology
flowchart TB
subgraph AWS["Customer AWS Account"]
direction TB
Internet(["Internet"])
Slack(["Slack workspace"])
subgraph Surfaces["QM surfaces"]
Portal["Portal & Web UI"]
Admin["Admin panel"]
end
Core["Core API — control plane"]
subgraph Data["Durable state"]
RDS[("RDS Postgres — sessions, memory, queue")]
S3[("S3 — snapshots and blobs")]
end
subgraph Sandboxes["Per-scope agent sandboxes"]
MV1["Lambda MicroVM — personal scope"]
MV2["Lambda MicroVM — room / channel"]
end
ECR["ECR — sandbox and core images"]
subgraph Platform["Platform & ops"]
SM["Secrets Manager"]
CW["CloudWatch"]
TF["Terraform state — S3 + DynamoDB lock"]
end
end
Internet --> Portal
Internet --> Admin
Slack --> Core
Portal --> Core
Admin --> Core
Core --> RDS
Core --> S3
Core --> MV1
Core --> MV2
MV1 --> S3
MV2 --> S3
Core -. images .-> ECR
MV1 -. images .-> ECR
MV2 -. images .-> ECR
Core --> SM
Core --> CW
Digitize managed operations
flowchart LR
subgraph Customer["Customer"]
Team["Your team"]
Acct["AWS account owner"]
end
subgraph Digitize["Digitize LLC"]
Op["QM operators"]
end
subgraph AWSAccount["Your AWS account"]
Role["IAM role — DigitizeQMOperator"]
Stack["QM stack — Terraform + ECR + RDS + S3 + MicroVMs"]
end
Team -->|"Web UI / Slack"| Stack
Acct -->|"Creates scoped role + ExternalId"| Role
Op -->|"AssumeRole — least privilege"| Role
Role -->|"terraform apply, qm up, monitoring"| Stack
IAM role, deployment repo, auth and model choices
Terraform apply, ECR images, secrets push
Live checks, admin chat, agent-computer verification
Bot in workspace, OAuth connectors as scoped
Team onboarding doc, operate schedule, escalation path
Transparent service fees. AWS and model API costs bill directly to your accounts. Estimate your year-one savings with the DIY vs managed cost calculator.
One-time deployment
Typical 5 business days
Managed ongoing ops · requires Launch
Flat rate — monitoring, upgrades, and incident response
Pilot offer: first 3 customers — Launch at $3,000 including 60 days Operate.
Estimated AWS infra: $370–870/mo for ~20 seats (varies by usage). LLM spend is separate (BYOK).
Adjust team size, engineer rate, and AWS assumptions in the
cost calculator.
Larger teams or compliance requirements? Contact us for a custom quote.
No. We deploy into your AWS account. You retain data ownership, network control, and CloudTrail visibility. We operate via a scoped IAM role — not multi-tenant SaaS.
Any region where QM's AWS backend is available, including Lambda MicroVM support. We preflight with aws lambda-microvms list-microvm-images. Default recommendation: us-west-2 or us-east-1 unless you have compliance requirements.
You do, via your own Anthropic, OpenAI, or OpenRouter keys (bring your own key). We configure QM budget caps and help you set sane org-wide limits. We do not markup model tokens.
No. QM is an MIT-licensed open-source project from Y Combinator. Digitize is an independent consulting company that deploys and operates QM on AWS for customers.
QM also supports Fly.io. This page focuses on our AWS offering for teams with existing AWS footprints or compliance needs. Contact us to discuss Fly.io or a hybrid approach.
Book a free 30-minute discovery call. We'll qualify fit, estimate AWS costs, and outline a go-live date.