Think. Design. Simulate. Deploy.
The AI workspace for cloud infrastructure. Design AWS, Azure, and GCP architectures on a visual canvas, simulate up to 100M RPS of real traffic, see your exact monthly cost — and catch the bottleneck before your users do. Then deploy in one click.
“Build an AI customer support platform.”
Trusted by teams at
From idea to production
in four moves.
The standard workflow makes you deploy first and find out later. Pinpole compresses design, cost estimation, load testing, and architecture review into one session — before a single dollar is spent.
Describe what you want.
Describe your system in plain English. Pinpole’s AI Architect designs the full architecture automatically — services, networking, permissions, everything.
Edit visually.
Drag services, connect APIs, modify resources, and collaborate with your team in real time — on an infinite multi-cloud canvas.
- Drag & drop 400+ cloud services
- Live multiplayer cursors
- Every edit stays deploy-ready
Simulate before you deploy.
Run production-grade traffic simulations against your design. Predict bottlenecks, estimate cloud costs, and identify failures — before they ever reach production.
- Model RPS, latency & error rates
- Forecast monthly cloud spend
- Find the bottleneck before your users do
Deploy in one click.
Sync GitHub, generate Terraform, deploy directly to your cloud, and monitor the rollout — without leaving the canvas.
- Sync GitHub & generate Terraform
- Deploy directly to AWS, Azure, GCP
- Monitor & roll back safely
Four pillars.
One workspace.
Your diagram knows your cost model. Your cost model knows your traffic pattern. Your load test runs before the infrastructure exists. That's the point.
AI Infrastructure Architect
Upload architecture diagrams, generate infrastructure from prompts, get plain-English explanations — and convert whiteboard sketches into production systems.
Simulation Engine
The feature nothing else has. Model traffic patterns, latency, failures, throttling, scaling behavior and monthly cost — before a single resource exists.
Visual Cloud Canvas
Drag & drop across every major cloud on one infinite canvas.
One-Click Deploy
Terraform, CloudFormation, direct deploy, previews & instant rollback.
Templates
Production-ready starting points for every kind of system.
The bridge between cloud architecture
and deployment.
Everything shipped in the Pinpole dashboard today, organized as the workflow you actually follow — design on canvas, simulate what it costs with thousands of users, connect your cloud and deploy, then manage the live stack.
Visual architecture design on an infinite canvas. No cloud account required to start.
Drag services from the AWS, GCP, Azure, and Cloudflare catalogs — 13 categories from Compute to AI/ML, with provider-accurate icons and full-text search.
32 reference architectures — Stripe-like Payments, Netflix-like Streaming, RAG Chatbot, Kubernetes Platform — preview, then replace or merge into your canvas.
Describe a workload in plain English and watch nodes and edges stream onto the canvas. Or upload a diagram photo — vision AI extracts services, connections, and groups.
Invalid source→target wiring is blocked per provider registry before it's created, with suggestions. Valid cross-cloud edges — like Cloudflare in front of AWS — are supported.
Click any node to configure VPC, region, instance type, runtime, and capacity — JSON-driven schemas per provider, with draft-before-save and per-node insights.
VPC-style group frames, text labels, zone boxes, and 18 architecture actors — users, mobile, SaaS, IoT, bank — for diagrams that read like the real system.
Autosave with revision tracking for solo work; team workspaces with role-based edit access and live canvas sync. Share read-only embeds without login.
60-step undo/redo, copy/paste, right-click context menus, Cmd+K command palette, auto-organize layout, and a guided first-run tour.
PNG for decks, JSON for tooling, animated GIF of simulations — and IaC as Terraform, CloudFormation, CDK, or ARM.
Run traffic through your design and watch cost and behaviour tick live — before spending on real infrastructure.
A 100ms loop updates every node and edge while the simulation runs. Pause, edit traffic config, and resume mid-run.
Logarithmic traffic slider with Constant, Ramp, Spike, and Wave patterns — or express load as total requests per day, week, or month.
Per-tick state for Lambda concurrency and cold starts, Kinesis shards, DynamoDB capacity, and SQS queue depth — 30+ dedicated processors.
$/sec, $/hour, $/day, and $/month projections update as the run progresses, with a per-node breakdown sorted by load.
Bedrock and AgentCore modelled with token and unit pricing — your GenAI stack costed like any other service.
Load %, sparklines, latency, and health per node; animated edges with throughput labels; P50/P95/P99 across the graph; alerts with actionable fixes.
A rule engine plus AI pass suggests cost, latency, and resilience improvements — one click applies the fix to the canvas and wires the edges.
Every run persists with config, peak RPS, cost, and recommendations. Share tokenized interactive simulation embeds — no login needed.
AI agents run batch simulations through the API and get an inline simulation canvas back in chat.
Go from canvas design to live infrastructure in your own cloud account — with zero long-lived keys stored.
Your cloud trusts Pinpole via standard OIDC/WIF federation. Every deploy uses short-lived credentials minted on demand — no stored secrets.
One-click CloudFormation Quick Create sets up the IAM OIDC provider and deploy role. Canvas → CloudFormation → AssumeRoleWithWebIdentity → live stack.
Provision the WIF pool via Terraform ZIP + Cloud Shell or Google OAuth, then deploy through Cloud Build in your own project.
Connect with an API token and apply Workers, Pages, KV, D1, Queues, R2, and DNS in real time — or import existing resources onto the canvas.
Generate an ARM template from the canvas and deploy through the Azure Portal link. Full OIDC connect is next on the roadmap.
No credentials at all: get the generated template plus a console Quick Create URL and run it from your own cloud console.
Link a repo, branch, and handler path to any Lambda node. Pinpole discovers SAM/Serverless functions and auto-syncs code on every push — no stack redeploy.
Terraform, CloudFormation, CDK, and SAM for AWS; Terraform and ARM for Azure; Terraform for GCP — with optional AI refinement of the output.
Shared AWS/GCP connections at team level, with role-gated deploy permissions and plan-based deploy quotas.
The canvas stays in charge after the deploy — sync changes, detect drift, and manage the live stack.
After connect, one button pushes saved canvas changes to the live stack — CloudFormation on AWS, Cloud Build on GCP, realtime on Cloudflare.
A diff engine compares canvas vs last deploy and routes small changes to direct API updates instead of full stack updates.
Push a single service's config — or one field — to a live AWS resource without redeploying anything else.
Compare the canvas against live CloudFormation and Lambda state on AWS, the last deployed snapshot on GCP, or resource fingerprints on Cloudflare.
Versioned per-workspace history with status, trigger, resources, and cost — plus live status polling, timeline, stack outputs, and console deep links.
Physical name, ARN, and endpoint URL per node, with "Open in AWS Console" links and a resource info API.
Per-deployment event streams, in-app toasts and inbox, and email preferences for deploy and simulation outcomes.
Role-based collaboration (owner, admin, dev, viewer), an immutable MFA-gated audit log, API/MCP tokens, and session management.
A CLI-style surface to query simulated state and explore quotas without leaving the session.
One workspace, every provider. Pinpole speaks your stack’s language.
Connect your entire stack.
Pinpole plugs into the tools you already use — code, clouds, chat, and observability. Set up in minutes, stay in sync forever.
An API-first platform.
Automate everything.
Everything in the UI,
available as an API.
Create architectures, run simulations, and trigger deployments programmatically. Type-safe SDKs for TypeScript and Python, webhooks for everything else.
- Architectures, simulations & deployments as resources
- Webhooks for deploy & simulation events
- Fine-grained API keys with scoped permissions
$ curl -X POST https://api.pinpole.cloud/v1/architectures \
-H "Authorization: Bearer $PINPOLE_API_KEY" \
-d '{
"prompt": "Event-driven image pipeline on AWS",
"simulate": { "rps": 1000 }
}'
// 201 Created
{
"id": "arch_8fk2m",
"resources": 27,
"simulation": { "p50_ms": 26, "monthly_cost": 182 }
}import { Pinpole } from "@pinpole/sdk";
const pinpole = new Pinpole();
const arch = await pinpole.architectures.create({
prompt: "Event-driven image pipeline on AWS",
});
const sim = await arch.simulate({ rps: 1_000 });
console.log(sim.monthlyCost); // $182
console.log(sim.bottlenecks); // []
await arch.deploy({ env: "production" });$ pinpole export --format terraform
✓ Generated 27 resources
├─ main.tf
├─ variables.tf
├─ networking.tf
└─ outputs.tf
$ terraform plan
Plan: 27 to add, 0 to change, 0 to destroy.Your infrastructure,
available to every AI agent.
The Pinpole MCP server exposes your canvases, simulations and deployments as tools any MCP-compatible agent can call — Claude, Cursor, or your own.
$ npm install -g @pinpole/mcp
✓ pinpole-mcp connected (9 tools)
› Design a RAG pipeline and estimate its cost
⏺ pinpole_build_architecture("RAG pipeline")
└─ 14 resources on canvas
⏺ pinpole_simulate_cost(rps: 500)
└─ $96/month · no bottlenecks
⏺ Done — canvas ready to review.Bring any model
into your workflows.
Choose which intelligence powers your AI Architect and agent workflows. Swap models per project, per workflow — or bring your own keys.
- Frontier models from every major lab
- BYO API keys — your data never trains anyone’s model
- Per-workflow routing & fallbacks
ClaudeAnthropic
GPT-5OpenAI
GeminiGoogle
LlamaMeta
MistralMistral AIReal teams. Real results.
Engineering teams use Pinpole to find failures, validate choices, and avoid costly mistakes — before a single resource is provisioned.
Traffic spike survived with zero downtime
A 45-minute simulation found two critical failures before 200,000 users hit the endpoint.
Read case study →First AWS bill vs. simulation prediction
A fintech team projected serverless spend to within 4% — and found $690/mo in optimizations pre-deploy.
Read case study →Monthly cost avoided on database selection
Simulation revealed the cost gap between DynamoDB and Aurora at their exact traffic profile.
Read analysis →Simulation that found 2 critical failures
See how pre-deployment modelling replaces guesswork with evidence before launch day.
All case studies →Less guessing. More shipping.
Infrastructure decisions made on guesswork are a systems failure, not a personal one. Pinpole gives engineers the data to make decisions they can defend.
Design
Visual cloud architecture that stays perfectly in sync with what actually runs.
Validate
Run production traffic simulations and know your costs and limits before launch.
Deploy
Ship with confidence — one click from validated design to running infrastructure.
What would you like to build?
Type it. Pinpole architects it.
Stop guessing.
Start building infrastructure
that works.
Free forever for individuals · No credit card required