Skip to content

Commit 2662848

Browse files
ViktarStarastsenkaViktarStarastsenka
authored
Create ai_assistant.md
1 parent 8366183 commit 2662848

File tree

1 file changed

+202
-0
lines changed

1 file changed

+202
-0
lines changed

‎src/vss/ai_assistant.md‎

Lines changed: 202 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,202 @@
1+
This tutorial demonstrates how to build an AI assistant's memory system with Redis as its memory core.
2+
3+
**Note**: Requires [Redis 8](https://hub.docker.com/_/redis/tags) for `HSETEX`, which adds per-field TTL for hashes - ideal for rate limiting to ensure fair resource usage.
4+
5+
### Architecture Overview
6+
| Layer | Description | Data type |
7+
| ---------- | ---------- | ---------- |
8+
| `Session History`| `Recent conversation context` | List |
9+
| `Rate Limiting` | `Per-user request throttling` | Hash |
10+
| `User Memory` | `Long-term facts and preferences` | Hash |
11+
12+
### Session History
13+
AI assistants need context from previous messages to provide coherent responses. Without conversation history, each interaction would be isolated. Redis lists are simple, ordered, and efficient for storing chat transcripts.
14+
15+
```redis:[run_confirmation=true] Store conversation history
16+
// Add user message to session
17+
LPUSH user:alice:history:session_001 '{"type": "human", "content": "What is the weather like?", "timestamp": 1717935001}'
18+
19+
// Add AI response
20+
LPUSH user:alice:history:session_001 '{"type": "ai", "content": "It is sunny with 75°F temperature.", "timestamp": 1717935002}'
21+
22+
// Add another user message
23+
LPUSH user:alice:history:session_001 '{"type": "human", "content": "Should I bring an umbrella?", "timestamp": 1717935003}'
24+
25+
// Add AI response
26+
LPUSH user:alice:history:session_001 '{"type": "ai", "content": "No umbrella needed today!", "timestamp": 1717935004}'
27+
```
28+
### Reading Conversation History
29+
You can retrieve recent messages to provide context to the AI.
30+
31+
```redis:[run_confirmation=no] Read conversation history
32+
// Get the 5 most recent messages
33+
LRANGE user:alice:history:session_001 0 4
34+
35+
// Get the full session
36+
LRANGE user:alice:history:session_001 0 -1
37+
```
38+
You may want to limit the size of history to retain only the N most recent items. Use LTRIM:
39+
```redis:[run_confirmation=yes] Read conversation history
40+
LTRIM user:alice:history:session_001 0 29 // keep only latest 30 items
41+
```
42+
43+
### Session Expiration
44+
Without expiration, session history will accumulate indefinitely. Expiring keys improves memory usage and ensures privacy.
45+
46+
```redis:[run_confirmation=true] Session expiration
47+
// Set session to expire in 24 hours
48+
EXPIRE user:alice:history:session_001 86400
49+
50+
// Set session to expire in 1 hour
51+
EXPIRE user:alice:history:session_001 3600
52+
53+
// Check remaining TTL
54+
TTL user:alice:history:session_001
55+
56+
// Remove expiration (make persistent)
57+
PERSIST user:alice:history:session_001
58+
```
59+
60+
### Rate Limiting
61+
Rate limiting prevents abuse and ensures fair usage across users. Redis hashes with field-level TTL via `HSETEX` are ideal for this.
62+
63+
```redis:[run_confirmation=true] Initialize Rate Limiting
64+
// On first request - set counter with 1-minute TTL
65+
HSETEX user:alice:rate_limit EX 60 FIELDS 1 requests_per_minute 1
66+
```
67+
68+
The `HINCR` command allows you to atomically increment the counter, preventing race conditions in high-concurrency scenarios.
69+
70+
```redis:[run_confirmation=true] Increment Requests
71+
// Increment request counter
72+
HINCRBY user:alice:rate_limit requests_per_minute 1
73+
74+
// Check if field exists and get count
75+
HEXISTS user:alice:rate_limit requests_per_minute
76+
HGET user:alice:rate_limit requests_per_minute
77+
78+
// Check TTL on the hash
79+
TTL user:alice:rate_limit
80+
```
81+
**Optionally**: if the count exceeds the allowed threshold, deny the operation.
82+
83+
Different time windows serve different purposes - per-minute prevents burst attacks, per-hour prevents sustained abuse, per-day enforces usage quotas.
84+
```redis:[run_confirmation=true] Rate Limiting with Different Time Windows
85+
// Set multiple rate limits with different TTLs
86+
HSETEX user:alice:rate_limit EX 60 FIELDS 2 requests_per_minute 1 requests_per_hour 1
87+
88+
// Daily rate limit (24 hours)
89+
HSETEX user:alice:rate_limit EX 86400 FIELDS 1 requests_per_day 1
90+
91+
// Check all rate limits
92+
HGETALL user:alice:rate_limit
93+
```
94+
95+
### User Memory (Persistent Preferences)
96+
AI assistants become more helpful when they remember user preferences, schedules, or relevant facts. This persistent memory enables personalization over time.
97+
98+
```redis:[run_confirmation=true] Store User Preferences
99+
// Always secure sensitive data using encryption at rest, access control (Redis ACLs), and comply with data protection laws (e.g., GDPR).
100+
// These values are stored with embeddings to support semantic recall later using vector search.
101+
HSET user:alice:pref:001 user_id "alice" content "prefers morning appointments before 10 AM" importance 9 timestamp 1717935000 embedding "\x40\x20\x00\x00\x3f\x80\x00\x00\x40\x40\x00\x00\x3f\x40\x00\x00\x40\x60\x00\x00\x40\x00\x00\x00\x3f\x00\x00\x00\x40\x80\x00\x00"
102+
103+
// Storing communication preference
104+
HSET user:alice:pref:002 user_id "alice" content "likes detailed explanations with examples" importance 8 timestamp 1717935000 embedding "\x3f\x40\x00\x00\x40\x60\x00\x00\x40\x00\x00\x00\x40\x20\x00\x00\x3f\x80\x00\x00\x40\x40\x00\x00\x40\x80\x00\x00\x3f\x00\x00\x00"
105+
106+
// Store work schedule preference
107+
HSET user:alice:pref:003 user_id "alice" content "works remotely on Fridays" importance 7 timestamp 1717935000 embedding "\x40\x80\x00\x00\x3f\x00\x00\x00\x40\x40\x00\x00\x40\x20\x00\x00\x3f\x80\x00\x00\x40\x60\x00\x00\x40\x00\x00\x00\x3f\x40\x00\x00"
108+
109+
// Store health information
110+
HSET user:alice:personal:001 user_id "alice" content "allergic to shellfish and nuts" importance 10 timestamp 1717935000 embedding "\x40\x00\x00\x00\x40\x20\x00\x00\x3f\x80\x00\x00\x40\x40\x00\x00\x40\x60\x00\x00\x3f\x40\x00\x00\x40\x80\x00\x00\x3f\x00\x00\x00"
111+
112+
// Store pet information
113+
HSET user:alice:personal:002 user_id "alice" content "has a golden retriever named Max, 3 years old" importance 7 timestamp 1717935000 embedding "\x3f\x80\x00\x00\x40\x40\x00\x00\x40\x00\x00\x00\x3f\x40\x00\x00\x40\x80\x00\x00\x40\x20\x00\x00\x3f\x00\x00\x00\x40\x60\x00\x00"
114+
115+
// Store family information
116+
HSET user:alice:personal:003 user_id "alice" content "married to Bob, two kids Sarah (8) and Tom (5)" importance 9 timestamp 1717935000 embedding "\x40\x60\x00\x00\x40\x00\x00\x00\x3f\x40\x00\x00\x40\x80\x00\x00\x40\x20\x00\x00\x3f\x80\x00\x00\x40\x40\x00\x00\x3f\x00\x00\x00"
117+
```
118+
119+
### Vector Search: Semantic Memory Recall
120+
Semantic search allows AI to retrieve relevant memory even when exact keywords don't match. For example, a query about "meetings" might return facts about "morning appointments."
121+
122+
Indexing persistent memory (User Memory) for semantically meaningful search.
123+
124+
```redis:[run_confirmation=true] Create a Vector Index
125+
// Create index for semantic search
126+
FT.CREATE idx:preferences
127+
ON HASH
128+
PREFIX 1 user:
129+
SCHEMA
130+
user_id TAG
131+
content TEXT
132+
importance NUMERIC
133+
timestamp NUMERIC
134+
embedding VECTOR HNSW 6
135+
TYPE FLOAT32
136+
DIM 8 // DIM 8 is only for demonstration purposes. Real embeddings are typically 128–1536 dimensions depending on the model (e.g., sentence-transformers).
137+
DISTANCE_METRIC COSINE
138+
```
139+
140+
### Search for most relevant memory entries
141+
142+
```redis:[run_confirmation=false] Find Top 5 Most Relevant User Memory Items
143+
FT.SEARCH idx:preferences
144+
"(@user_id:{alice}) => [KNN 5 @embedding $vec AS score]"
145+
PARAMS 2 vec "\x40\x00\x00\x00\x40\x20\x00\x00\x3f\x80\x00\x00\x40\x40\x00\x00\x40\x60\x00\x00\x3f\x40\x00\x00\x40\x80\x00\x00\x3f\x00\x00\x00"
146+
RETURN 4 content importance score timestamp
147+
SORTBY score ASC
148+
DIALECT 2
149+
```
150+
151+
```redis:[run_confirmation=false] Search High-Importance Items Only
152+
FT.SEARCH idx:preferences
153+
"(@user_id:{alice} @importance:[8 +inf]) => [KNN 3 @embedding $vec AS score]"
154+
PARAMS 2 vec "\x40\x40\x00\x00\x3f\x00\x00\x00\x40\x80\x00\x00\x40\x20\x00\x00\x3f\x80\x00\x00\x40\x60\x00\x00\x40\x00\x00\x00\x3f\x40\x00\x00"
155+
RETURN 4 content importance score timestamp
156+
SORTBY score ASC
157+
DIALECT 2
158+
```
159+
```redis:[run_confirmation=false] Search Recent User Memories
160+
FT.SEARCH idx:preferences
161+
"(@user_id:{alice} @timestamp:[1717935000 +inf]) => [KNN 5 @embedding $vec AS score]"
162+
PARAMS 2 vec "\x3f\x80\x00\x00\x40\x40\x00\x00\x40\x00\x00\x00\x3f\x40\x00\x00\x40\x80\x00\x00\x40\x20\x00\x00\x3f\x00\x00\x00\x40\x60\x00\x00"
163+
RETURN 3 content timestamp score
164+
SORTBY score ASC
165+
DIALECT 2
166+
```
167+
168+
### Memory State Monitoring
169+
Understanding what's stored in memory helps debug issues, optimize performance, and ensure data quality. It's also essential for user privacy compliance.
170+
```redis:[run_confirmation=false] Monitor user sessions
171+
// Get approximate memory usage of session
172+
MEMORY USAGE user:alice:history:session_001
173+
174+
// Get session statistics
175+
LLEN user:alice:history:session_001
176+
TTL user:alice:history:session_001
177+
```
178+
### Data Cleanup
179+
Remove all data related to a user (e.g., for GDPR compliance).
180+
181+
```redis:[run_confirmation=true] Delete user data
182+
// Remove all user data (GDPR compliance)
183+
DEL user:alice:history:session_001
184+
DEL user:alice:history:session_002
185+
DEL user:alice:rate_limit
186+
DEL user:alice:pref:001
187+
DEL user:alice:pref:002
188+
DEL user:alice:pref:003
189+
DEL user:alice:personal:001
190+
DEL user:alice:personal:002
191+
DEL user:alice:personal:003
192+
DEL user:alice:work:001
193+
DEL user:alice:work:002
194+
DEL user:alice:work:003
195+
```
196+
197+
### Next Steps
198+
Now that your assistant has memory and meaning, you can:
199+
- Combine with RAG Pipelines
200+
- Use sentence-transformers to generate high-dimensional vectors
201+
- Add [Redis Flex](https://redis.io/solutions/flex/?utm_source=redisinsight&utm_medium=app&utm_campaign=tutorials) for fallback persistence
202+
- Use Redis ACLs to isolate users, enforce quotas, and monitor usage

0 commit comments

Comments
 (0)