Aerostack
Aerostack
AI Infrastructure features are coming soon — the content below shows our vision for this product.
//AI Infrastructure

Ship your AI product.
We handle the infrastructure.

Stream to users, bring any model, RAG on your data — with auth, billing, and WebSocket scale built in. One SDK.

AI Streaming
BYOM / Any Model
Vector DB
Edge Scale
AI MODELAuthBillingStreamingGatewayMetricsUsage
//The Problem

6 things every AI product needs that aren't the AI.

What you needTime to build
User auth + API key management3–4 weeks
Vector DB + RAG pipeline3–4 weeks
Real-time streaming (WS / SSE)2–3 weeks
Billing + subscription tiers4–6 weeks
Usage metering + limits2 weeks
AI Gateway routing + fallbacks2–3 weeks
Total4–5 months before your product ships

Aerostack ships all of it. So you can ship yours.

Before Aerostack / After Aerostack

Interactive Evolution

Infrastructure Redefined.

Drag the slider to see how Aerostack unifies chaotic manual provisioning into a single, type-safe TypeScript interface.

auto_fix_highAerostack Abstraction
infrastructure.ts
1import { Aerostack, Stack } from '@aerostack/sdk';
2 
3// Define your cloud stack in pure TypeScript
4const app = new Aerostack('payment-service');
5 
6await app.deploy({
7  database: 'redis-cluster',
8  queue: 'kafka-stream',
9  storage: 's3-bucket',
10  security: 'iam-strict'
11});
12 
13console.log("🚀 Infrastructure ready!");
warningManual Architecture
dns
REDIS_MASTER_01_PROD
Status: Connection Lost
storage
S3_LEGACY_CONTENT_BUCKET
report_problem502 Bad Gateway
router
VPC
terraform {
  Backend: "S3"
}
scheduleSSL Expired (2d)
compare_arrows
//Streaming

Your users see responses as they stream.

Drop in the built-in chat widget, or stream directly into your own UI with the SDK. WebSocket available now. SSE in beta.

sub-100ms TTFT30K concurrent300+ PoPs
index.html
<!-- Zero code. One script tag. -->
<script
  src="https://cdn.aerostack.dev/widget.js"
  data-project="my-app"
  data-model="gpt-4o"
  data-theme="dark"
></script>
//Architecture

One platform between your model
and your users.

Your ModelBYOM · any endpointBYOMAerostack ModelsGPT-4o · Claude · LlamaAI GatewayRouteAuthTrackMeterFallbackVector DBRAG + EmbeddingsStream / ChatWebSocket · SSEAuthJWT · OAuth · KeysUsageMeters per user
BYOM

Your Model / Endpoint

Any API — OpenAI-compatible, HuggingFace, custom

Aerostack Models

GPT-4o · Claude · Gemini · Llama — one API key

// One SDK
AI GatewayVector DBAuthBillingWebSocketUsage

Your Users

Scales to 30,000+ concurrent WebSocket connections

//Features

What your team gets
out of the box.

Auth

Sign-up, login, API key management for your users.

Billing

Per-token, per-call, or flat subscription billing.

Subscriptions

Tiered plans, free trials, upgrades and downgrades.

Chat Widget

Drop-in streaming chat or stream to your own UI.

AI Gateway

Route to any model, handle fallbacks automatically.

Vector DB

Built-in embeddings + semantic search for RAG apps.

AI Model Access

GPT-4o, Claude, Gemini, Llama — one API key.

Real-time

WebSocket + SSE scaling to 30K+ concurrent.

Usage Dashboard

Per-user consumption tracking in real-time.

//Edge Infrastructure

Built to scale from day one.

300+ global edge PoPs. WebSocket connections backed by Durable Objects. No cold starts, no load balancers.

300+

Edge locations

30K+

Concurrent WebSockets

<50ms

Global p95 latency

AI MODELAuthBillingStreamingGatewayMetricsUsage
//Use Cases

Built for teams shipping
enterprise AI products.

Enterprise Copilot

Internal AI over company data — HR, legal, finance. Private deployment. BYOM.

Customer Support AI

Deflect tickets. Train on your docs. Stream answers live with the built-in widget.

AI Dev Tools

Code review bots, IDE plugins, PR analysis — with per-seat billing built in.

AI Search

Semantic search over products, docs, or knowledge bases using Vector DB.

AI API Products

Sell your model as an API — auth, billing, metering, and a gateway all included.

Content Platform

Blog generators, repurposing tools — multi-tenant with per-user billing.

//Integration

10 lines. That's your entire
AI backend.

import { Aerostack } from '@aerostack/sdk';

const ai = new Aerostack({ slug: 'my-app' });

const stream = await ai.stream({
  model: 'gpt-4o',          // or bring your own endpoint
  messages: [{ role: 'user', content: prompt }],
  userId: session.userId,   // auto-metered
});

for await (const chunk of stream) {
  res.write(chunk.text);
}
Project Templates

Deploy in seconds,
not days.

10+ pre-built RAG pipelines for every use case. Pick a template, upload your data, and go live.

📚
Document Intelligence

Docs Q&A Bot

Upload your docs — users get accurate answers with citations.

RAG + LLM
One-click deploy →
🎧
Customer Support

Customer Support Bot

Train on your FAQs — deflect tickets automatically.

RAG + LLM
One-click deploy →
⚖️
Legal

Legal Document Analyzer

Search contracts, extract clauses, identify risks.

RAG + 70B LLM
One-click deploy →
🔍
Discovery

Semantic Search

Natural language search over products, docs, or knowledge bases.

RAG + LLM
One-click deploy →
🔍
Developer Tools

Code Review Assistant

Review PRs against your team's coding standards.

RAG + 70B LLM
One-click deploy →
🏢
HR & Ops

Internal Knowledge Base

Give your team instant answers from your company wiki.

RAG + LLM
One-click deploy →
Meeting SummarizerResume ScreenerCompetitor IntelligenceSemantic Product Search+ more

All templates include: RAG pipeline · Vector DB · Auth · Metered billing

Available now — start building below
//get started

Build skills. Deploy globally.
Start earning.

Universal skills, MCP workspace, edge functions, marketplace — one platform, generous free tier.