mirror of
https://github.com/supermemoryai/supermemory.git
synced 2026-05-18 23:36:00 +00:00
feat: profile page updates
This commit is contained in:
Naman Bansal
parent
35ac9e086b
commit
01a09fac56
5 changed files with 802 additions and 2 deletions
|
|
@ -4,7 +4,7 @@ description: "Use Supermemory with Vercel AI SDK for seamless memory management"
|
|||
sidebarTitle: "Overview"
|
||||
---
|
||||
|
||||
The Supermemory AI SDK provides native integration with Vercel's AI SDK through two approaches: **Memory Tools** for agent-based interactions and **Infinite Chat** for automatic context management.
|
||||
The Supermemory AI SDK provides native integration with Vercel's AI SDK through three approaches: **User Profiles** for automatic personalization, **Memory Tools** for agent-based interactions, and **Infinite Chat** for automatic context management.
|
||||
|
||||
<Card title="Supermemory tools on npm" icon="npm" href="https://www.npmjs.com/package/@supermemory/tools">
|
||||
Check out the NPM page for more details
|
||||
|
|
@ -16,6 +16,25 @@ The Supermemory AI SDK provides native integration with Vercel's AI SDK through
|
|||
npm install @supermemory/tools
|
||||
```
|
||||
|
||||
## User Profiles with Middleware
|
||||
|
||||
Automatically inject user profiles into every LLM call for instant personalization.
|
||||
|
||||
```typescript
|
||||
import { generateText } from "ai"
|
||||
import { withSupermemory } from "@supermemory/tools/ai-sdk"
|
||||
import { openai } from "@ai-sdk/openai"
|
||||
|
||||
// Wrap your model with Supermemory - profiles are automatically injected
|
||||
const modelWithMemory = withSupermemory(openai("gpt-4"), "user-123")
|
||||
|
||||
const result = await generateText({
|
||||
model: modelWithMemory,
|
||||
messages: [{ role: "user", content: "What do you know about me?" }]
|
||||
})
|
||||
// The model automatically has the user's profile context!
|
||||
```
|
||||
|
||||
## Memory Tools
|
||||
|
||||
Add memory capabilities to AI agents with search, add, and fetch operations.
|
||||
|
|
@ -64,12 +83,17 @@ const result = await streamText({
|
|||
|
||||
| Approach | Use Case |
|
||||
|----------|----------|
|
||||
| User Profiles | Personalized LLM responses with automatic user context |
|
||||
| Memory Tools | AI agents that need explicit memory control |
|
||||
| Infinite Chat | Chat applications with automatic context |
|
||||
|
||||
## Next Steps
|
||||
|
||||
<CardGroup cols={2}>
|
||||
<CardGroup cols={3}>
|
||||
<Card title="User Profiles" icon="user" href="/ai-sdk/user-profiles">
|
||||
Automatic personalization with profiles
|
||||
</Card>
|
||||
|
||||
<Card title="Memory Tools" icon="wrench" href="/ai-sdk/memory-tools">
|
||||
Agent-based memory management
|
||||
</Card>
|
||||
|
|
|
|||
204
apps/docs/ai-sdk/user-profiles.mdx
Normal file
204
apps/docs/ai-sdk/user-profiles.mdx
Normal file
|
|
@ -0,0 +1,204 @@
|
|||
---
|
||||
title: "User Profiles with AI SDK"
|
||||
description: "Automatically inject user profiles into LLM calls for instant personalization"
|
||||
sidebarTitle: "User Profiles"
|
||||
---
|
||||
|
||||
## Overview
|
||||
|
||||
The `withSupermemory` middleware automatically injects user profiles into your LLM calls, providing instant personalization without manual prompt engineering or API calls.
|
||||
|
||||
<Note>
|
||||
**New to User Profiles?** Read the [conceptual overview](/user-profiles) to understand what profiles are and why they're powerful for LLM personalization.
|
||||
</Note>
|
||||
|
||||
## Quick Start
|
||||
|
||||
```typescript
|
||||
import { generateText } from "ai"
|
||||
import { withSupermemory } from "@supermemory/tools/ai-sdk"
|
||||
import { openai } from "@ai-sdk/openai"
|
||||
|
||||
// Wrap any model with Supermemory middleware
|
||||
const modelWithMemory = withSupermemory(
|
||||
openai("gpt-4"), // Your base model
|
||||
"user-123" // Container tag (user ID)
|
||||
)
|
||||
|
||||
// Use normally - profiles are automatically injected!
|
||||
const result = await generateText({
|
||||
model: modelWithMemory,
|
||||
messages: [{ role: "user", content: "Help me with my current project" }]
|
||||
})
|
||||
|
||||
// The model knows about the user's background, skills, and current work!
|
||||
```
|
||||
|
||||
## How It Works
|
||||
|
||||
The `withSupermemory` middleware:
|
||||
|
||||
1. **Intercepts** your LLM calls before they reach the model
|
||||
2. **Fetches** the user's profile based on the container tag
|
||||
3. **Injects** profile data into the system prompt automatically
|
||||
4. **Forwards** the enhanced prompt to your LLM
|
||||
|
||||
All of this happens transparently - you write code as if using a normal model, but get personalized responses.
|
||||
|
||||
## Memory Search Modes
|
||||
|
||||
Configure how the middleware retrieves and uses memory:
|
||||
|
||||
### Profile Mode (Default)
|
||||
|
||||
Retrieves the user's complete profile without query-specific search. Best for general personalization.
|
||||
|
||||
```typescript
|
||||
// Default behavior - profile mode
|
||||
const model = withSupermemory(openai("gpt-4"), "user-123")
|
||||
|
||||
// Or explicitly specify
|
||||
const model = withSupermemory(openai("gpt-4"), "user-123", {
|
||||
mode: "profile"
|
||||
})
|
||||
|
||||
const result = await generateText({
|
||||
model,
|
||||
messages: [{ role: "user", content: "What do you know about me?" }]
|
||||
})
|
||||
// Response uses full user profile for context
|
||||
```
|
||||
|
||||
### Query Mode
|
||||
|
||||
Searches memories based on the user's specific message. Best for finding relevant information.
|
||||
|
||||
```typescript
|
||||
const model = withSupermemory(openai("gpt-4"), "user-123", {
|
||||
mode: "query"
|
||||
})
|
||||
|
||||
const result = await generateText({
|
||||
model,
|
||||
messages: [{
|
||||
role: "user",
|
||||
content: "What was that Python script I wrote last week?"
|
||||
}]
|
||||
})
|
||||
// Searches for memories about Python scripts from last week
|
||||
```
|
||||
|
||||
### Full Mode
|
||||
|
||||
Combines profile AND query-based search for comprehensive context. Best for complex interactions.
|
||||
|
||||
```typescript
|
||||
const model = withSupermemory(openai("gpt-4"), "user-123", {
|
||||
mode: "full"
|
||||
})
|
||||
|
||||
const result = await generateText({
|
||||
model,
|
||||
messages: [{
|
||||
role: "user",
|
||||
content: "Help me debug this similar to what we did before"
|
||||
}]
|
||||
})
|
||||
// Uses both profile (user's expertise) AND search (previous debugging sessions)
|
||||
```
|
||||
|
||||
## Verbose Logging
|
||||
|
||||
Enable detailed logging to see exactly what's happening:
|
||||
|
||||
```typescript
|
||||
const model = withSupermemory(openai("gpt-4"), "user-123", {
|
||||
verbose: true // Enable detailed logging
|
||||
})
|
||||
|
||||
const result = await generateText({
|
||||
model,
|
||||
messages: [{ role: "user", content: "Where do I live?" }]
|
||||
})
|
||||
|
||||
// Console output:
|
||||
// [supermemory] Searching memories for container: user-123
|
||||
// [supermemory] User message: Where do I live?
|
||||
// [supermemory] System prompt exists: false
|
||||
// [supermemory] Found 3 memories
|
||||
// [supermemory] Memory content: You live in San Francisco, California...
|
||||
// [supermemory] Creating new system prompt with memories
|
||||
```
|
||||
|
||||
## Comparison with Direct API
|
||||
|
||||
The AI SDK middleware abstracts away the complexity of manual profile management:
|
||||
|
||||
<Tabs>
|
||||
<Tab title="With AI SDK (Simple)">
|
||||
```typescript
|
||||
// One line setup
|
||||
const model = withSupermemory(openai("gpt-4"), "user-123")
|
||||
|
||||
// Use normally
|
||||
const result = await generateText({
|
||||
model,
|
||||
messages: [{ role: "user", content: "Help me" }]
|
||||
})
|
||||
```
|
||||
</Tab>
|
||||
|
||||
<Tab title="Without AI SDK (Complex)">
|
||||
```typescript
|
||||
// Manual profile fetching
|
||||
const profileRes = await fetch('https://api.supermemory.ai/v4/profile', {
|
||||
method: 'POST',
|
||||
headers: { /* ... */ },
|
||||
body: JSON.stringify({ containerTag: "user-123" })
|
||||
})
|
||||
const profile = await profileRes.json()
|
||||
|
||||
// Manual prompt construction
|
||||
const systemPrompt = `User Profile:\n${profile.profile.static?.join('\n')}`
|
||||
|
||||
// Manual LLM call with profile
|
||||
const result = await generateText({
|
||||
model: openai("gpt-4"),
|
||||
messages: [
|
||||
{ role: "system", content: systemPrompt },
|
||||
{ role: "user", content: "Help me" }
|
||||
]
|
||||
})
|
||||
```
|
||||
</Tab>
|
||||
</Tabs>
|
||||
|
||||
## Limitations
|
||||
|
||||
- **Beta Feature**: The `withSupermemory` middleware is currently in beta
|
||||
- **Container Tag Required**: You must provide a valid container tag
|
||||
- **API Key Required**: Ensure `SUPERMEMORY_API_KEY` is set in your environment
|
||||
|
||||
## Next Steps
|
||||
|
||||
<CardGroup cols={2}>
|
||||
<Card title="User Profiles Concepts" icon="brain" href="/user-profiles">
|
||||
Understand how profiles work conceptually
|
||||
</Card>
|
||||
|
||||
<Card title="Memory Tools" icon="wrench" href="/ai-sdk/memory-tools">
|
||||
Add explicit memory operations to your agents
|
||||
</Card>
|
||||
|
||||
<Card title="API Reference" icon="code" href="https://api.supermemory.ai/v3/reference#tag/profile">
|
||||
Explore the underlying profile API
|
||||
</Card>
|
||||
|
||||
<Card title="NPM Package" icon="npm" href="https://www.npmjs.com/package/@supermemory/tools">
|
||||
View the package on NPM
|
||||
</Card>
|
||||
</CardGroup>
|
||||
|
||||
<Info>
|
||||
**Pro Tip**: Start with profile mode for general personalization, then experiment with query and full modes as you understand your use case better.
|
||||
</Info>
|
||||
|
|
@ -112,6 +112,7 @@
|
|||
]
|
||||
},
|
||||
"search/filtering",
|
||||
"user-profiles",
|
||||
"memory-api/track-progress",
|
||||
{
|
||||
"group": "List Memories",
|
||||
|
|
@ -193,6 +194,7 @@
|
|||
"icon": "triangle",
|
||||
"pages": [
|
||||
"ai-sdk/overview",
|
||||
"ai-sdk/user-profiles",
|
||||
"ai-sdk/memory-tools",
|
||||
"ai-sdk/infinite-chat",
|
||||
"ai-sdk/npm"
|
||||
|
|
|
|||
BIN
apps/docs/images/static-dynamic-profile.png
Normal file
BIN
apps/docs/images/static-dynamic-profile.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 154 KiB |
570
apps/docs/user-profiles.mdx
Normal file
570
apps/docs/user-profiles.mdx
Normal file
|
|
@ -0,0 +1,570 @@
|
|||
---
|
||||
title: "User Profiles - Persistent Context for LLMs"
|
||||
description: "Automatically maintained user profiles that provide instant, comprehensive context to your LLMs"
|
||||
sidebarTitle: "User Profiles"
|
||||
icon: "user"
|
||||
---
|
||||
|
||||
## What are User Profiles?
|
||||
|
||||
User profiles are **automatically maintained collections of facts about your users** that Supermemory builds from all their interactions and content. Think of it as a persistent "about me" document that's always up-to-date and instantly accessible.
|
||||
|
||||
Instead of searching through memories every time you need context about a user, profiles give you:
|
||||
- **Instant access** to comprehensive user information
|
||||
- **Automatic updates** as users interact with your system
|
||||
- **Two-tier structure** separating permanent facts from temporary context
|
||||
|
||||
<Note>
|
||||
Profile data can be appended to the system prompt so that it's always sent to your LLM and you don't need to run multiple queries.
|
||||
</Note>
|
||||
|
||||
## Static vs Dynamic Profiles
|
||||
|
||||
|
||||
|
||||
Profiles are intelligently divided into two categories:
|
||||
|
||||
### Static Profile
|
||||
**Long-term, stable facts that define who the user is**
|
||||
|
||||
These are facts that rarely change - the foundational information about a user that remains consistent over time.
|
||||
|
||||
Examples:
|
||||
- "Sarah Chen is a senior software engineer at TechCorp"
|
||||
- "Sarah specializes in distributed systems and Kubernetes"
|
||||
- "Sarah has a PhD in Computer Science from MIT"
|
||||
- "Sarah prefers technical documentation over video tutorials"
|
||||
|
||||
### Dynamic Profile
|
||||
**Recent context and temporary information**
|
||||
|
||||
These are current activities, recent interests, and temporary states that provide immediate context.
|
||||
|
||||
Examples:
|
||||
- "Sarah is currently migrating the payment service to microservices"
|
||||
- "Sarah recently started learning Rust for a side project"
|
||||
- "Sarah is preparing for a conference talk next month"
|
||||
- "Sarah is debugging a memory leak in the authentication service"
|
||||
|
||||
<Accordion title="How are profiles different from search?" defaultOpen>
|
||||
**Traditional Search**: You query "What does Sarah know about Kubernetes?" and get specific memory chunks about Kubernetes.
|
||||
|
||||
**User Profiles**: You get Sarah's complete professional context instantly - her role, expertise, preferences, and current projects - without needing to craft specific queries.
|
||||
|
||||
The profile is **always there**, providing consistent personalization across every interaction.
|
||||
</Accordion>
|
||||
|
||||
## Why We Built Profiles
|
||||
|
||||
### The Problem with Search-Only Approaches
|
||||
|
||||
Traditional memory systems rely entirely on search, which has fundamental limitations:
|
||||
|
||||
1. **Search is too narrow**: When you search for "project updates", you miss that the user prefers bullet points, works in PST timezone, and uses specific technical terminology.
|
||||
|
||||
2. **Search is repetitive**: Every chat message triggers multiple searches for basic context that rarely changes.
|
||||
|
||||
3. **Search misses relationships**: Individual memory chunks don't capture the full picture of who someone is and how different facts relate.
|
||||
|
||||
|
||||
Profiles solve these problems by maintaining a **persistent, holistic view** of each user:
|
||||
## How Profiles Work with Search
|
||||
|
||||
Profiles don't replace search - they complement it perfectly:
|
||||
|
||||
<Steps>
|
||||
<Step title="Profile provides foundation">
|
||||
The user's profile gives your LLM comprehensive background context about who they are, what they know, and what they're working on.
|
||||
</Step>
|
||||
|
||||
<Step title="Search adds specificity">
|
||||
When you need specific information (like "error in deployment yesterday"), search finds those exact memories.
|
||||
</Step>
|
||||
|
||||
<Step title="Combined for perfect context">
|
||||
Your LLM gets both the broad understanding from profiles AND the specific details from search.
|
||||
</Step>
|
||||
</Steps>
|
||||
|
||||
### Real-World Example
|
||||
|
||||
Imagine a user asks: **"Can you help me debug this?"**
|
||||
|
||||
**Without profiles**: The LLM has no context about the user's expertise level, current projects, or debugging preferences.
|
||||
|
||||
**With profiles**: The LLM knows:
|
||||
- The user is a senior engineer (adjust technical level)
|
||||
- They're working on a payment service migration (likely context)
|
||||
- They prefer command-line tools over GUIs (tool suggestions)
|
||||
- They recently had issues with memory leaks (possible connection)
|
||||
|
||||
## Technical Implementation
|
||||
|
||||
### Endpoint Details
|
||||
|
||||
Based on the [API reference](https://api.supermemory.ai/v3/reference#tag/profile), the profile endpoint provides a simple interface:
|
||||
|
||||
**Endpoint**: `POST /v4/profile`
|
||||
|
||||
### Request Parameters
|
||||
|
||||
| Parameter | Type | Required | Description |
|
||||
|-----------|------|----------|-------------|
|
||||
| `containerTag` | string | **Yes** | The container tag (usually user ID) to get profiles for |
|
||||
| `q` | string | No | Optional search query to include search results with the profile |
|
||||
|
||||
### Response Structure
|
||||
|
||||
The response includes both profile data and optional search results:
|
||||
|
||||
```json
|
||||
{
|
||||
"profile": {
|
||||
"static": [
|
||||
"User is a software engineer",
|
||||
"User specializes in Python and React"
|
||||
],
|
||||
"dynamic": [
|
||||
"User is working on Project Alpha",
|
||||
"User recently started learning Rust"
|
||||
]
|
||||
},
|
||||
"searchResults": {
|
||||
"results": [...], // Only if 'q' parameter was provided
|
||||
"total": 15,
|
||||
"timing": 45.2
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
## Code Examples
|
||||
|
||||
### Basic Profile Retrieval
|
||||
|
||||
<CodeGroup>
|
||||
|
||||
```typescript TypeScript
|
||||
// Direct API call using fetch
|
||||
const response = await fetch('https://api.supermemory.ai/v4/profile', {
|
||||
method: 'POST',
|
||||
headers: {
|
||||
'Authorization': `Bearer ${process.env.SUPERMEMORY_API_KEY}`,
|
||||
'Content-Type': 'application/json'
|
||||
},
|
||||
body: JSON.stringify({
|
||||
containerTag: 'user_123'
|
||||
})
|
||||
});
|
||||
|
||||
const data = await response.json();
|
||||
|
||||
console.log("Static facts:", data.profile.static);
|
||||
console.log("Dynamic context:", data.profile.dynamic);
|
||||
|
||||
// Use in your LLM prompt
|
||||
const systemPrompt = `
|
||||
User Context:
|
||||
${data.profile.static?.join('\n') || ''}
|
||||
|
||||
Current Activity:
|
||||
${data.profile.dynamic?.join('\n') || ''}
|
||||
|
||||
Please provide personalized assistance based on this context.
|
||||
`;
|
||||
```
|
||||
|
||||
```python Python
|
||||
import requests
|
||||
import os
|
||||
|
||||
# Direct API call
|
||||
response = requests.post(
|
||||
'https://api.supermemory.ai/v4/profile',
|
||||
headers={
|
||||
'Authorization': f'Bearer {os.getenv("SUPERMEMORY_API_KEY")}',
|
||||
'Content-Type': 'application/json'
|
||||
},
|
||||
json={
|
||||
'containerTag': 'user_123'
|
||||
}
|
||||
)
|
||||
|
||||
data = response.json()
|
||||
|
||||
print("Static facts:", data['profile']['static'])
|
||||
print("Dynamic context:", data['profile']['dynamic'])
|
||||
|
||||
# Use in your LLM prompt
|
||||
static_context = '\n'.join(data['profile'].get('static', []))
|
||||
dynamic_context = '\n'.join(data['profile'].get('dynamic', []))
|
||||
|
||||
system_prompt = f"""
|
||||
User Context:
|
||||
{static_context}
|
||||
|
||||
Current Activity:
|
||||
{dynamic_context}
|
||||
|
||||
Please provide personalized assistance based on this context.
|
||||
"""
|
||||
```
|
||||
|
||||
```bash cURL
|
||||
curl -X POST https://api.supermemory.ai/v4/profile \
|
||||
-H "Authorization: Bearer YOUR_API_KEY" \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"containerTag": "user_123"
|
||||
}'
|
||||
```
|
||||
|
||||
</CodeGroup>
|
||||
|
||||
### Profile with Search
|
||||
|
||||
Sometimes you want both the user's profile AND specific search results:
|
||||
|
||||
<CodeGroup>
|
||||
|
||||
```typescript TypeScript
|
||||
// Get profile with search results
|
||||
const response = await fetch('https://api.supermemory.ai/v4/profile', {
|
||||
method: 'POST',
|
||||
headers: {
|
||||
'Authorization': `Bearer ${process.env.SUPERMEMORY_API_KEY}`,
|
||||
'Content-Type': 'application/json'
|
||||
},
|
||||
body: JSON.stringify({
|
||||
containerTag: 'user_123',
|
||||
q: 'deployment errors yesterday' // Optional search query
|
||||
})
|
||||
});
|
||||
|
||||
const data = await response.json();
|
||||
|
||||
// Now you have both profile and specific search results
|
||||
const profile = data.profile;
|
||||
const searchResults = data.searchResults?.results || [];
|
||||
|
||||
// Combine for comprehensive context
|
||||
const context = {
|
||||
userBackground: profile.static,
|
||||
currentContext: profile.dynamic,
|
||||
specificInfo: searchResults.map(r => r.content)
|
||||
};
|
||||
```
|
||||
|
||||
```python Python
|
||||
import requests
|
||||
|
||||
# Get profile with search results
|
||||
response = requests.post(
|
||||
'https://api.supermemory.ai/v4/profile',
|
||||
headers={
|
||||
'Authorization': f'Bearer {os.getenv("SUPERMEMORY_API_KEY")}',
|
||||
'Content-Type': 'application/json'
|
||||
},
|
||||
json={
|
||||
'containerTag': 'user_123',
|
||||
'q': 'deployment errors yesterday' # Optional search query
|
||||
}
|
||||
)
|
||||
|
||||
data = response.json()
|
||||
|
||||
# Access both profile and search results
|
||||
profile = data['profile']
|
||||
search_results = data.get('searchResults', {}).get('results', [])
|
||||
|
||||
# Combine for comprehensive context
|
||||
context = {
|
||||
'user_background': profile.get('static', []),
|
||||
'current_context': profile.get('dynamic', []),
|
||||
'specific_info': [r['content'] for r in search_results]
|
||||
}
|
||||
```
|
||||
|
||||
</CodeGroup>
|
||||
|
||||
### Integration with Chat Applications
|
||||
|
||||
Here's how to use profiles in a real chat application:
|
||||
|
||||
<CodeGroup>
|
||||
|
||||
```typescript TypeScript
|
||||
async function handleChatMessage(userId: string, message: string) {
|
||||
// Get user profile for personalization
|
||||
const profileResponse = await fetch('https://api.supermemory.ai/v4/profile', {
|
||||
method: 'POST',
|
||||
headers: {
|
||||
'Authorization': `Bearer ${process.env.SUPERMEMORY_API_KEY}`,
|
||||
'Content-Type': 'application/json'
|
||||
},
|
||||
body: JSON.stringify({
|
||||
containerTag: userId
|
||||
})
|
||||
});
|
||||
|
||||
const profileData = await profileResponse.json();
|
||||
|
||||
// Build personalized system prompt
|
||||
const systemPrompt = buildPersonalizedPrompt(profileData.profile);
|
||||
|
||||
// Send to your LLM with context
|
||||
const response = await llm.chat({
|
||||
messages: [
|
||||
{ role: "system", content: systemPrompt },
|
||||
{ role: "user", content: message }
|
||||
]
|
||||
});
|
||||
|
||||
return response;
|
||||
}
|
||||
|
||||
function buildPersonalizedPrompt(profile: any) {
|
||||
return `You are assisting a user with the following context:
|
||||
|
||||
ABOUT THE USER:
|
||||
${profile.static?.join('\n') || 'No profile information yet.'}
|
||||
|
||||
CURRENT CONTEXT:
|
||||
${profile.dynamic?.join('\n') || 'No recent activity.'}
|
||||
|
||||
Provide responses that are personalized to their expertise level,
|
||||
preferences, and current work context.`;
|
||||
}
|
||||
```
|
||||
|
||||
```python Python
|
||||
import requests
|
||||
import os
|
||||
|
||||
async def handle_chat_message(user_id: str, message: str):
|
||||
# Get user profile for personalization
|
||||
response = requests.post(
|
||||
'https://api.supermemory.ai/v4/profile',
|
||||
headers={
|
||||
'Authorization': f'Bearer {os.getenv("SUPERMEMORY_API_KEY")}',
|
||||
'Content-Type': 'application/json'
|
||||
},
|
||||
json={'containerTag': user_id}
|
||||
)
|
||||
|
||||
profile_data = response.json()
|
||||
|
||||
# Build personalized system prompt
|
||||
system_prompt = build_personalized_prompt(profile_data['profile'])
|
||||
|
||||
# Send to your LLM with context
|
||||
llm_response = await llm.chat(
|
||||
messages=[
|
||||
{"role": "system", "content": system_prompt},
|
||||
{"role": "user", "content": message}
|
||||
]
|
||||
)
|
||||
|
||||
return llm_response
|
||||
|
||||
def build_personalized_prompt(profile):
|
||||
static_facts = '\n'.join(profile.get('static', ['No profile information yet.']))
|
||||
dynamic_context = '\n'.join(profile.get('dynamic', ['No recent activity.']))
|
||||
|
||||
return f"""You are assisting a user with the following context:
|
||||
|
||||
ABOUT THE USER:
|
||||
{static_facts}
|
||||
|
||||
CURRENT CONTEXT:
|
||||
{dynamic_context}
|
||||
|
||||
Provide responses that are personalized to their expertise level,
|
||||
preferences, and current work context."""
|
||||
```
|
||||
|
||||
</CodeGroup>
|
||||
|
||||
## AI SDK Integration
|
||||
|
||||
<Note>
|
||||
The Supermemory AI SDK provides a more elegant way to use profiles through the `withSupermemory` middleware, which automatically handles profile retrieval and injection into your LLM prompts.
|
||||
</Note>
|
||||
|
||||
### Automatic Profile Integration
|
||||
|
||||
The AI SDK's `withSupermemory` middleware abstracts away all the profile endpoint complexity:
|
||||
|
||||
```typescript
|
||||
import { generateText } from "ai"
|
||||
import { withSupermemory } from "@supermemory/tools/ai-sdk"
|
||||
import { openai } from "@ai-sdk/openai"
|
||||
|
||||
// Automatically injects user profile into every LLM call
|
||||
const modelWithMemory = withSupermemory(openai("gpt-4"), "user_123")
|
||||
|
||||
const result = await generateText({
|
||||
model: modelWithMemory,
|
||||
messages: [{ role: "user", content: "What do you know about me?" }],
|
||||
})
|
||||
|
||||
// The model automatically has access to the user's profile!
|
||||
```
|
||||
|
||||
### Memory Search Modes
|
||||
|
||||
The AI SDK supports three modes for memory retrieval:
|
||||
|
||||
#### Profile Mode (Default)
|
||||
Retrieves user profile memories without query filtering:
|
||||
|
||||
```typescript
|
||||
import { generateText } from "ai"
|
||||
import { withSupermemory } from "@supermemory/tools/ai-sdk"
|
||||
import { openai } from "@ai-sdk/openai"
|
||||
|
||||
// Uses profile mode by default - gets all user profile memories
|
||||
const modelWithMemory = withSupermemory(openai("gpt-4"), "user-123")
|
||||
|
||||
// Explicitly specify profile mode
|
||||
const modelWithProfile = withSupermemory(openai("gpt-4"), "user-123", {
|
||||
mode: "profile"
|
||||
})
|
||||
|
||||
const result = await generateText({
|
||||
model: modelWithMemory,
|
||||
messages: [{ role: "user", content: "What do you know about me?" }],
|
||||
})
|
||||
```
|
||||
|
||||
#### Query Mode
|
||||
Searches memories based on the user's message:
|
||||
|
||||
```typescript
|
||||
import { generateText } from "ai"
|
||||
import { withSupermemory } from "@supermemory/tools/ai-sdk"
|
||||
import { openai } from "@ai-sdk/openai"
|
||||
|
||||
const modelWithQuery = withSupermemory(openai("gpt-4"), "user-123", {
|
||||
mode: "query"
|
||||
})
|
||||
|
||||
const result = await generateText({
|
||||
model: modelWithQuery,
|
||||
messages: [{ role: "user", content: "What's my favorite programming language?" }],
|
||||
})
|
||||
```
|
||||
|
||||
#### Full Mode
|
||||
Combines both profile and query results:
|
||||
|
||||
```typescript
|
||||
import { generateText } from "ai"
|
||||
import { withSupermemory } from "@supermemory/tools/ai-sdk"
|
||||
import { openai } from "@ai-sdk/openai"
|
||||
|
||||
const modelWithFull = withSupermemory(openai("gpt-4"), "user-123", {
|
||||
mode: "full"
|
||||
})
|
||||
|
||||
const result = await generateText({
|
||||
model: modelWithFull,
|
||||
messages: [{ role: "user", content: "Tell me about my preferences" }],
|
||||
})
|
||||
```
|
||||
|
||||
<Card title="Learn More About AI SDK" icon="triangle" href="/ai-sdk/overview">
|
||||
Explore the full capabilities of the Supermemory AI SDK, including tools for adding memories, searching, and automatic profile injection.
|
||||
</Card>
|
||||
|
||||
### Understanding the Modes (Without AI SDK)
|
||||
|
||||
When using the API directly without the AI SDK:
|
||||
|
||||
- **Profile Only**: Call `/v4/profile` and add the profile data to your system prompt. This gives persistent user context without query-specific search.
|
||||
|
||||
- **Query Only**: Use the `/v4/search` endpoint with the user's specific question to find relevant memories based on their current query. Read [the search docs.](/search/overview)
|
||||
|
||||
- **Full Mode**: Combine both approaches - add profile data to the system prompt AND use the search endpoint for conversational context based on the user's specific query. This provides the most comprehensive context.
|
||||
|
||||
```typescript
|
||||
// Full mode example without AI SDK
|
||||
async function getFullContext(userId: string, userQuery: string) {
|
||||
// 1. Get user profile for system prompt
|
||||
const profileResponse = await fetch('https://api.supermemory.ai/v4/profile', {
|
||||
method: 'POST',
|
||||
headers: { /* ... */ },
|
||||
body: JSON.stringify({ containerTag: userId })
|
||||
});
|
||||
const profileData = await profileResponse.json();
|
||||
|
||||
// 2. Search for query-specific memories
|
||||
const searchResponse = await fetch('https://api.supermemory.ai/v3/search', {
|
||||
method: 'POST',
|
||||
headers: { /* ... */ },
|
||||
body: JSON.stringify({
|
||||
q: userQuery,
|
||||
containerTag: userId
|
||||
})
|
||||
});
|
||||
const searchData = await searchResponse.json();
|
||||
|
||||
// 3. Combine both in your prompt
|
||||
return {
|
||||
systemPrompt: `User Profile:\n${profileData.profile.static?.join('\n')}`,
|
||||
queryContext: searchData.results
|
||||
};
|
||||
}
|
||||
```
|
||||
Or you can also juse use the `q` parameter in the `v4/profiles` endpoint to get those search results. I just wanted to demonstrate how you can use search and profile separately, so I put this elaborate code snippet.
|
||||
|
||||
## How Profiles are Built
|
||||
|
||||
Profiles are **automatically constructed and maintained** through Supermemory's ingestion pipeline:
|
||||
|
||||
<Steps>
|
||||
<Step title="Content Ingestion">
|
||||
When users add documents, chat, or any content to Supermemory, it goes through the standard ingestion workflow.
|
||||
</Step>
|
||||
|
||||
<Step title="Intelligence Extraction">
|
||||
AI analyzes the content to extract not just memories, but also facts about the user themselves.
|
||||
</Step>
|
||||
|
||||
<Step title="Profile Operations">
|
||||
The system generates profile operations (add, update, or remove facts) based on the new information.
|
||||
</Step>
|
||||
|
||||
<Step title="Automatic Updates">
|
||||
Profiles are updated in real-time, ensuring they always reflect the latest information about the user.
|
||||
</Step>
|
||||
</Steps>
|
||||
|
||||
<Note>
|
||||
You don't need to manually manage profiles - they're automatically maintained as users interact with your system. Just ingest content normally, and profiles build themselves.
|
||||
</Note>
|
||||
|
||||
|
||||
## Common Use Cases
|
||||
|
||||
### Personalized AI Assistants
|
||||
Profiles ensure your AI assistant remembers user preferences, expertise, and context across conversations.
|
||||
|
||||
### Customer Support Systems
|
||||
Support agents (or AI) instantly see customer history, preferences, and current issues without manual searches.
|
||||
|
||||
### Educational Platforms
|
||||
Adapt content difficulty and teaching style based on the learner's profile and progress.
|
||||
|
||||
### Development Tools
|
||||
IDE assistants that understand your coding style, current projects, and technical preferences.
|
||||
|
||||
## Performance Benefits
|
||||
|
||||
Profiles provide significant performance improvements:
|
||||
|
||||
| Metric | Without Profiles | With Profiles |
|
||||
|--------|-----------------|---------------|
|
||||
| Context Retrieval | 3-5 search queries | 1 profile call |
|
||||
| Response Time | 200-500ms | 50-100ms |
|
||||
| Token Usage | High (multiple searches) | Low (single response) |
|
||||
| Consistency | Varies by search quality | Always comprehensive |
|
||||
Loading…
Add table
Add a link
Reference in a new issue