Serverless Postgres, done right

The first multi-tenant, fully compatible, open-source Postgres backend for agents, developers, and builders who vibe. Deploy extremely cost-optimized and reliable micro DBs with consistent performance that autoscale. Postgres architected with true multi-tenancy over object storage to support high availability, instant provisioning and millions of 1¢ DBs with 20-50% lower compute cost.

⚡ QUICK_SPECS
[1]1¢ databases with $0 idle cost and no cold starts
[2]20-50% less compute cost. Pay for only CPU-ms used
[3]Clone, time travel and snapshot instantly
[4]Move to provisioned compute with no downtime in seconds
Postgres DBCreating...
Interact below
Claim ID (24h)Provisioning…

$0 when idle

Grap: Idle vs Usage

No charges when your database isn't being used. Unlike other "serverless" solutions, we mean it—zero dollars for idle time.

1¢ databases

Vintage Stamp Illustration

Create databases for pennies. No minimum commitment, no reserved capacity. Start small, scale infinitely.

Pay per CPU-ms

Architecture Diagram

Precise billing by CPU milliseconds. No rounding up to hours or minutes. Pay for exactly what you use, nothing more.

TECHNICAL NOTE

Our multi-tenant architecture shares compute resources across databases while maintaining complete isolation and security. This allows us to offer true serverless pricing without the traditional overhead costs of dedicated instances.

Built for developer and agent workflows

Developer-first features designed for modern workflows. Fork, time travel, preview, and provision databases instantly—no waiting, no complexity.

FIG 1.4.1 — INSTANT FORK
production_db
fork in 0.2s
dev_feature_a
test_env
agent_workspace

Instant fork

Clone entire databases in milliseconds. Create isolated environments for testing, development, or AI agents without copying data.

→ Copy-on-write technology
FIG 1.4.2 — TIME TRAVEL
NOW
2 hours ago
Restore to pre-migration state
1 day ago
Query historical data
1 week ago
Recover deleted records

Time travel

Query any point in your database's history. Restore to previous states, recover data, or analyze changes over time.

→ Point-in-time recovery
FIG 1.4.3 — PREVIEWS
GIT COMMIT
feat: add user profiles
⚡ Auto-detected migration
PREVIEW DATABASE
preview-pr-1234.nile.dev
app-preview-1234.vercel.app
✓ Live

Preview environments

Automatic preview databases for every pull request. Test schema changes and migrations in isolation before merging.

→ GitHub/GitLab integration
FIG 1.4.4 — PROVISIONING
Traditional
15-30 min
Other Serverless
2-5 min
Nile
< 1 sec
Request received
0.1s
Database allocated
0.3s
Connection ready
0.5s

Instant provisioning

New databases ready in under a second. No waiting, no setup wizard, no configuration. Just instant, production-ready Postgres.

→ Sub-second cold starts
🤖 AGENT WORKFLOWS

Built for autonomous agents

AI agents can programmatically create, fork, and manage databases without human intervention. Perfect for multi-tenant SaaS, per-user databases, or dynamic environments that scale with your agent workflows.

API-FIRST
Full REST & GraphQL APIs
ISOLATION
Per-agent DB instances
SCALE
Thousands of DBs instantly

Provisioned compute

When you need dedicated resources, switch to provisioned compute instantly. Zero downtime migrations, vertical scaling, and enhanced isolation—all on demand.

FIG 1.5.1 — SEAMLESS MIGRATION
SERVERLESS MODE
CPUAuto
MemoryAuto
Shared Resources
$0.01/db + CPU-ms
ONE-CLICK SWITCH
MIGRATION STATUS
Hot standby ready
0.2s
Traffic switched
0.3s
Zero dropped queries
0.0
PROVISIONED MODE
CPU4 cores
Memory16 GB
Dedicated Resources
$0.15/hour fixed
DOWNTIME
0ms
Seamless transition
MIGRATION TIME
~1 sec
Hot standby ready
REVERSIBLE
Yes
Switch back anytime
FIG 1.5.2 — VERTICAL SCALING
Small
2 cores
8 GB
Medium
4 cores
16 GB
Large
8 cores
32 GB
XLarge
16 cores
64 GB
SCALING ACTIONS
Instant scaling • No connection drops • Live migration

Vertical scaling

Scale compute resources up or down instantly as your needs change. From small workloads to enterprise scale, adjust on the fly with zero downtime.

→ Live resize without restarts
FIG 1.5.3 — ISOLATION
SERVERLESS (SHARED)
1
2
3
4
5
6
7
8
Multiple tenants • Shared resources
UPGRADE TO PROVISIONED
PROVISIONED (DEDICATED)
1
Single tenant • Dedicated resources • Full isolation
ISOLATION BENEFITS
Predictable performance
No noisy neighbor issues
Enhanced security boundaries
Compliance requirements

Enhanced isolation

Move from shared multi-tenant to dedicated single-tenant infrastructure. Get predictable performance and complete resource isolation for production workloads.

→ Dedicated compute & memory
WHEN TO USE PROVISIONED
High traffic

Production apps with consistent high load benefit from dedicated resources.

Compliance

Meet regulatory requirements that mandate dedicated infrastructure.

Predictability

Guaranteed performance without the variability of shared resources.

FIG 1.2.1 — TECHNICAL REFERENCE

Architecture Overview

Understanding Postgres internal architecture with dual backend processes, complete isolation, query processing pipelines, and shared resource management.

PostgreSQL Architecture Diagram showing dual backend processes with parser, analyzer, rewriter, planner, and executor components, along with shared buffer pool and storage manager
NOTE_01: PROCESS ISOLATION

Each database connection spawns a dedicated backend process (FIG 1.0, FIG 2.0) with isolated memory management, ensuring complete independence and fault tolerance.

NOTE_02: SHARED RESOURCES

Shared buffer pool, WAL buffer, and commit log are managed by the unified Storage Manager, providing efficient resource utilization across all connections.

Postgres backend to build and scale your applications faster

More than just a database. Nile comes with built-in authentication, vector embeddings, hosted MCP server, and AI agent capabilities—everything you need to ship faster.

FIG 1.6.1 — AUTHENTICATION
👤
CLIENT REQUEST
login@example.com
NILE AUTH LAYER
Validate
Sessions
Tokens
MFA
JWT Token • expires: 7d
✓ Authenticated

Optional auth

Drop-in authentication with user management, sessions, JWT tokens, and MFA. Skip weeks of auth implementation and security concerns.

→ OAuth, SSO, social logins
FIG 1.6.2 — VECTOR EMBEDDINGS
QUERY VECTOR
0.8
0.5
0.9
0.3
0.7
0.9
0.1
0.6
pgvector • cosine similarity
Document A
94%
Document B
87%
Document C
81%

Vector embeddings

Native pgvector support for semantic search, RAG, and AI applications. Store and query embeddings at scale without separate vector databases.

→ Optimized for similarity search
FIG 1.6.3 — HOSTED MCP
AI CLIENT (CLAUDE, GPT, ETC.)
🤖 Claude
🤖 GPT-4
🤖 Gemini
MCP Protocol
NILE MCP SERVER
Available Tools:
• query_database(sql)
• manage_users(action)
• vector_search(query)
🗄️
POSTGRES DATABASE
production_db

Hosted MCP

Model Context Protocol server hosted and managed for you. Connect AI agents to your database with pre-built tools and secure access controls.

→ Zero MCP server setup
FIG 1.6.4 — AI AGENT
USER PROMPT
"Show me users who signed up last week"
🤖
NILE AI AGENT
1.
Parse natural language → SQL
2.
Validate permissions & scope
3.
Execute query safely
4.
Format results for user
GENERATED SQL
SELECT * FROM users
WHERE created_at >= NOW() - INTERVAL '7 days'
ORDER BY created_at DESC;
RESULTS
Found 142 users • Avg: 20/day

AI agent

Built-in AI agent that understands natural language queries and translates them to SQL. Query your database, manage data, and build insights conversationally.

→ Natural language to SQL
INTEGRATED STACK
All-in-one

Database, auth, vectors, and AI in a single platform

Unified API

One SDK for all features with consistent patterns

Ship faster

Skip months of infrastructure setup and integration

Stay focused

Build your product, not your database backend

CALL TO ACTION

Start building today

Join developers building the next generation of multi-tenant applications.

NILE
© 2026 NILE ALL RIGHTS RESERVED