New 8k Context Window. 10GB of Memories. No problem.
Stop starting every chat from zero.
The universal memory layer for Ollama. Inject years of personal context into any local model with zero latency. 100% Sovereign.
Local-First. Privacy-Native. No Cloud.
INFO: Cross-referencing Obsidian Vault and Claude History...
LOAD: 3,420 semantic nodes active for 'Deion'.
> First night in LA. I'm feeling a bit out of place.
AI: That's the 'Pearson to LAX' transition, Deion. I've detected the shift in your metadata and prioritized your 'Career Transitions' and 'Adaptability' nodes in the current context stream. [Source: Claude/2024_Journal.json]
> Yeah, just a lot to handle. My twists are a mess from the flight too.
AI: I noticed you bookmarked 'Nappily Naturals' in Leimert Park while you were still in Toronto. Since you're 6ft tall, I've already checked—they have high-back styling chairs so you won't be hunched over during your touch-up. [Source: Obsidian/Physiology.md]
> Wait, how did you know about the chair thing? I never told you that.
AI: I cross-referenced your 'Physiology' note from your local Obsidian vault with the stylist's shop photos. Velqua doesn't just remember what you say—it remembers who you are.
01
Connect
Link Velqua to your Ollama endpoint. It sits transparently between your UI and your models.
02
Index
Drag in your Obsidian vault, ChatGPT exports, or Notion databases. Velqua builds a local semantic map.
03
Chat
Talk to any model. Velqua injects the right memories at the right time, automatically.
Connected Intelligence
Obsidian
Live local indexing
Claude & GPT
History ingestion
Notion
Workspace exports
Ollama
Native Proxy
Hardware-Aware Scaling
GPU VRAM
Memory Budget
Functional Window
8GB (Entry)
200 Tokens
Active Context Only
16GB (Mid)
500 Tokens
Balanced Recall
24GB (Pro)
1,000 Tokens
Full Biography Recall