Shared memory for regulated companies and their AI agents
High-assurance infrastructure for AI agents
Stage
Not given
Not given
Founded
Not given
Not given
Based in
US
US
Pricing model
Freemium
Open source
Pricing
Team plan: $10/person/month with 14-day free trial, includes meeting recording, email and calendar enrichment, knowledge organized by topic, private scopes and projects, agent keys with custom permissions, import existing brain, and MCP for all agents. Enterprise plan: custom pricing for regulated a
Not given
Key features
Records and transcribes meetings with visible participant
Automatically extracts facts from email and calendar
Scoped access control with automatic sorting
Private projects limited to members
Agent API keys with read-only or read-write permissions
Works with Claude, ChatGPT, Cursor, Codex over MCP
Import existing brain from markdown, Obsidian, or GBrain
Facts organized by topic with full history
Raw email stays private to owner, team sees extracted facts only
180 days of email history imported on day one
Harpe: chooses what an AI agent reads under a token budget with consistent results
Kibisis: keeps agent memory on your hardware with AES-256-GCM encryption
Argus: append-only record that detects edits with cryptographic verification
No network code by default for local and disconnected use
MIT License with published code, benchmarks and raw results
What makes it different
Every fact is scoped automatically by sensitivity level and agents see only what their teammate can access
Agents can write back to the same brain over MCP, updating shared knowledge in real time
CASA Tier 2 security assessed and verified by Google for Gmail and Calendar access
Onboard entire team in under 5 minutes with one-click Google sign-in
Data never used to train AI models, privacy verified in policy
Measured empirical validation across 6,111 questions outperforming full-context models
Zero plaintext on disk with encrypted vaults matching plaintext accuracy
78.3% fewer tokens sent to the model with a 1,200-token budget per turn
Preserves conversational chronology and relational qualifiers under token budgets
Nysa
Perseus Computing