Alchemyst AI — Vercel AI SDK Integration

Alchemyst AI is the context layer for your LLM applications — it remembers, reasons, and injects contextual intelligence automatically into every call. This package provides a seamless integration with Vercel’s AI SDK to enhance your Gen-AI apps with memory, retrieval, and context-aware toolchains — all through a single line of configuration.

🚀 Installation

npm install @alchemystai/aisdk
# or
yarn add @alchemystai/aisdk
#or
pnpm add @alchemystai/aisdk
# or
bun add @alchemystai/aisdk

⚡ Quick Start

Here’s how to plug Alchemyst AI into your ai SDK call in one line:

import { streamText } from 'ai';
import { alchemystTools } from '@alchemystai/aisdk';

const result = await streamText({
  model: "gpt-5-nano",
  prompt: "Remember that my name is Alice",
  tools: alchemystTools("YOUR_ALCHEMYST_AI_KEY", true, true)
});

This automatically attaches the Alchemyst Context Engine to your model call — enabling persistent memory and context-aware generation across sessions.

🧩 API Reference

alchemystTools(apiKey: string, enableMemory?: boolean, enableRetrieval?: boolean)

| Parameter | Type | Default | Description | | ----------------- | --------- | ------- | ------------------------------------------------------- | | apiKey | string | — | Your Alchemyst AI API key. Required. | | enableMemory | boolean | true | Enables contextual memory between user sessions. | | enableRetrieval | boolean | true | Enables semantic retrieval from your connected sources. |

Returns an object compatible with the tools parameter of the ai SDK functions (streamText, generateText, streamObject, etc.).

🧠 What It Does

Once integrated, Alchemyst AI automatically:

• Persists context across user sessions.
• Augments prompts with retrieved knowledge from your Alchemyst AI workspace.
• Supports custom tool functions for domain-specific reasoning.
• Runs entirely server-side — no extra infrastructure required.

💡 Example: Contextual Chatbot

import { streamText } from 'ai';
import { alchemystTools } from '@alchemystai/aisdk';

export async function POST(req: Request) {
  const { messages } = await req.json();

  const response = await streamText({
    model: 'gpt-5-nano',
    messages,
    tools: alchemystTools(process.env.ALCHEMYST_API_KEY!, true, true),
  });

  return response.toAIStreamResponse();
}

With this, your AI app will:

• Remember previous user messages (via Alchemyst context memory)
• Retrieve relevant knowledge chunks
• Enrich every prompt dynamically before sending it to the model

🔒 Environment Variables

Set your API key in your environment:

ALCHEMYST_API_KEY=sk-xxxxxx

Then reference it in your code as:

alchemystTools(process.env.ALCHEMYST_API_KEY!);

🧰 Supported SDK Methods

Alchemyst AI integrates with all Vercel AI SDK entry points:

| SDK Function | Supported | Notes | | ---------------- | --------- | ------------------------------- | | streamText | ✅ | Real-time streaming generation | | generateText | ✅ | Synchronous generation | | streamObject | ✅ | Structured JSON output | | generateObject | ✅ | Non-streaming structured output |

🧪 Example with Retrieval Off

const result = await streamText({
  model: "gpt-5-nano",
  prompt: "Hello there!",
  tools: alchemystTools("YOUR_ALCHEMYST_AI_KEY", true, false),
});

📜 License

MIT © 2025 Alchemyst AI