← Back to explorer
Example workload/workloads/platforms

Platforms

Let customers safely deploy custom logic on your product.

For programmable SaaS and platform businesses where each tenant, plugin, storefront, or generated app needs isolated execution, routing, limits, and platform-controlled extensibility.

programmable platformcustomer codetenant codepluginsextensibilityembedded automationwhite-label platformapp builder

Recommended primitives

Workers for Platforms

core

Managed multi-tenant runtime for platforms that deploy or execute customer code on Cloudflare.

Multi-tenant runtime for customer-deployed code

Workers

supporting

Serverless edge functions. The entry point for most Cloudflare apps.

Dispatch and platform-controlled orchestration

Durable Objects

supporting

Stateful single-threaded actors with strong consistency, storage, and WebSocket support.

Per-tenant coordination and shared state when needed

D1

optional

Managed serverless SQLite for relational app data and CRUD workloads under about 10 GB per database.

Platform metadata or tenant app data

KV

optional

Eventually consistent key-value storage for cache, sessions, config, and read-heavy lookups.

Routing tables and read-heavy configuration

Sandboxes

optional

Isolated execution environments for untrusted or user-submitted code.

Higher-isolation execution for richer generated environments

Architecture notes

  • A platform-controlled dispatch layer routes requests to tenant-owned code.
  • Strong tenant isolation, limits, and observability are first-class requirements.
  • Custom domains, bindings, and white-label experience are part of the product surface, not afterthoughts.

Starter prompts

We run a SaaS product and want customers to write custom workflow logic and API integrations.
Build a platform where each customer deploys its own storefront logic on our domain.
We are building an AI app builder where every generated app becomes its own isolated deployment.

Avoid notes

Treating a normal app like a platform

This workload only fits when customer extensibility or per-tenant deployable logic is central to the product.

Ad hoc tenant execution model

Routing, limits, domains, and isolation become hard to manage if customer code execution is bolted onto a standard app backend.

Consider: Use Workers for Platforms as the execution model and keep the product in control of policy.

Resources

Try your own version

Start from one of these prompts or describe your own variation on the homepage to get a primitive recommendation.

Open the explorer