Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Build reliable AI Workflows and Agents with humans in the loop, structured outputs and durable execution.
| Date | Stars |
|---|---|
| 2026-07-31 | 445 |
| 2026-08-02 | 445 |
| 2026-08-06 | 445 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
<div align="center">
<img src="./assets/logo.png" alt="Inferable Logo" width="200" />
# Inferable
Build reliable AI Workflows with humans in the loop
  
</div>
## What is Inferable?
Inferable is a managed durable execution runtime for creating AI workflows with humans in the loop. Create structured outputs from any LLM, ask humans for approval via Slack or Email, with versioned, long-running workflows for backwards compatibility.
## ⚡️ Quick Start
Follow the [quick start guide](https://docs.inferable.ai/pages/quick-start) to get started with Inferable.
## 🔑 Key Features
Here are some of the key features of Inferable.
### 📦 Workflows that execute in your own infrastructure
Workflows execute in your own infrastructure, even behind firewalls or private VPCs. No deployment step is required. We use long polling to connect to your infrastructure, so there is no need to open any inbound ports.
```typescript
const workflow = inferable.workflows.create({
name: "simple",
inputSchema: z.object({
executionId: z.string(),
greeting: z.string(),
}),
});
```
### 🔄 Versioned Workflows for backward compatibility
When you need to change the input schema or the logic of a workflow, you can create a new version of the workflow. Inferable will maintain version affinity for currently executing workflows, so you can roll out new versions gradually. See [Workflows](https://docs.inferable.ai/pages/workflows).
```typescript
workflow.version(1).define(async (ctx, input) => {
// ...
});
workflow.version(2).define(async (ctx, input) => {
// ...
});
```
### 🧑💼 Human-in-the-Loop with approval workflows
Inferable allows you to integrate human approval and intervention with full context preservation. See [Human-in-the-Loop](https://docs.inferable.ai/pages/human-in-the-loop).
```typescript
deleteUserWorkflow.version(1).define(async (ctx, input) => {
// ... existing workflow code ...
if (!ctx.approved) {
return Interrupt.approval({
message: `I need your approval to delete the user ${input.userId}. Is this ok?`,
destination: {
type: "email",
// The email address to notify
email: "[email protected]",
},
});
}
await db.customers.delete({
userId: input.userId,
});
});
```
### 🏗️ Structured Outputs with automatic parsing, validation, and retries
Inferable automatically parses and validates structured outputs, and retries failed executions. See [Structured Outputs](https://docs.inferable.ai/pages/structured-outputs).
```typescript
workflow.version(1).define(async (ctx, input) => {
const { ticketType } = ctx.llm.structured({
input: `Ticket text: ${input.ticketText}`,
schema: z.object({
ticketType: z.enum(["data-deletion", "refund", "other"]),
}),
});
// do something with the items
console.log(ticketType);
});
```
And more stuff...
- [Notifications](https://docs.inferable.ai/pages/notifications) to send notifications to users via Slack or Email.
- [Memoized Results](https://docs.inferable.ai/pages/results) to cache the results of side-effects and expensive operations in a distributed way.
- [Obervability](https://docs.inferable.ai/pages/observability) in a timeline view, or plug into your own observability tools.
- [Agents](https://docs.inferable.ai/pages/agents) to create agents that can use tools to achieve pre-defined goals.
- Developer-friendly SDKs in [Node.js](./sdk-node/README.md), and [Go](./sdk-go/README.md) supported with more languages coming Excerpt of 5,713 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:171d6c8e956dbaa2, topic:llm