Skip to main content
Use tier-0 pinned memories to store preferences that should always be recalled.

​
Store preferences

import requests

headers = {"Authorization": "Bearer vlx_your_key"}
url = "https://api.velixarai.com/v1/memory"

# Store multiple preferences
prefs = [
    "User prefers dark mode",
    "User wants concise responses, no fluff",
    "User's timezone is America/Los_Angeles",
    "User prefers Python over JavaScript",
]

for pref in prefs:
    requests.post(url, headers=headers, json={
        "content": pref,
        "user_id": "user_123",
        "tier": 0,          # Pinned — always recalled first
        "type": "preference"
    })

​
Recall at conversation start

# At the start of every conversation, load user preferences
resp = requests.get("https://api.velixarai.com/v1/memory/search",
    headers=headers,
    params={
        "q": "user preferences and settings",
        "user_id": "user_123",
        "tiers": "0",   # Only pinned memories
        "limit": 20
    })

prefs = [m["content"] for m in resp.json().get("memories", [])]
system_prompt = "User preferences:\n" + "\n".join(f"- {p}" for p in prefs)

​
Update when preferences change

When a user says “actually, switch me to light mode”: 
# Search for the old preference
old = requests.get("https://api.velixarai.com/v1/memory/search",
    headers=headers,
    params={"q": "dark mode light mode theme", "user_id": "user_123", "tiers": "0"}
).json()["memories"]

# Delete old, store new
if old:
    requests.delete(f"https://api.velixarai.com/v1/memory/{old[0]['id']}", headers=headers)

requests.post("https://api.velixarai.com/v1/memory", headers=headers, json={
    "content": "User prefers light mode",
    "user_id": "user_123",
    "tier": 0,
    "type": "preference"
})
With auto-extraction enabled, preference changes are detected and updated automatically — no manual code needed.