Now accepting early access signups

A mind for every
AI agent

Awareness, emotional intelligence, relationships, proactiveness, and deep memory for any AI agent. One API call.

# One command. Your agent gets a mind.
npx mind0 init --api-key m0_...

Free to start · Works with any agent · Any LLM provider

Built for OpenClaw Claude Code Cursor LangChain CrewAI LangChain Any Agent

Same agent. Different understanding.

Without Mind0
User: Draft an email to David
Agent: Sure! What would you like to say to David? Could you provide some context about the email?
With Mind0
User: Draft an email to David
Agent: David hasn't replied in 2 days and your board meeting is tomorrow. Here's a draft that references the term sheet discussion and suggests a quick call before the board meets. I've kept the tone direct — I know you're under pressure this week.

Same LLM. Same agent code. Mind0 provides the context that transforms generic responses into informed ones.

Everything your agent needs to understand its human

This is what your agent knows. Not after weeks of setup. After one call.

Structured JSON response — click to expand
emotional_intelligence anxious (0.35/0.70) · declining · calibrated
"current_state": { "valence": 0.35, "arousal": 0.70, "primary_emotion": "anxious", "confidence": 0.82 }, "baseline": { "calibrated": true, "data_points": 11, "avg_valence": 0.60, "avg_arousal": 0.50 }, "trajectory": { "7_day_trend": "declining" }, "crisis_assessment": { "tier": 0, "signals": [] }, "behavioral_signals": { "agency_score": 0.8, "hedging_language": "low", "session_frequency": "stable" }, "engagement_guidance": { "tone": "calm, direct", "approach": "Lead with single action", "lean_into": ["quick wins"], "avoid": ["long-term planning"] }
memories 2 relevant — "Fundraising Series A, targeting $3M"
[ { "memory": "Fundraising Series A, targeting $3M", "importance": 8, "sensitivity": "internal", "topics": ["fundraising", "series-a"] }, { "memory": "David at Sequoia introduced by Rachel", "importance": 7, "sensitivity": "internal", "topics": ["investors", "relationships"] } ]
personal_context Sarah Chen · Acme Corp · direct style
"identity": { "name": "Sarah Chen", "company": "Acme Corp", "timezone": "America/New_York" }, "communication_profile": { "message_style": "direct", "peak_activity_hours": [9, 10, 14] }, "critical_dates": { "board_meeting": "2026-04-01" }, "personal_rules": [ { "name": "no-early-meetings", "rule": "Never schedule before 10am" } ]
living_context "Sarah Chen — CEO, fundraising Series A..."
"synthesis": "Sarah Chen — CEO at Acme Corp, fundraising Series A targeting $3M. Solo founder, first startup. More stressed than usual this week. Board meeting tomorrow with key investors. David from Sequoia hasn't replied in 2 days.", "emotional_signal": "high stress, determined", "priority_context": [ "fundraise", "board prep" ], "confidence_score": 0.85
temporal Monday · March 31 · Afternoon · Workday
{ "date": "March 31, 2026", "day_of_week": "Monday", "time_of_day": "Afternoon", "workday": true }
email_awareness 2 need attention · 1 at-risk thread Gmail
"summary": { "needs_you": 2, "unread_count": 15, "inbox_health": "needs attention" }, "insights": [ { "type": "follow_up_risk", "priority": "high", "message": "David (Sequoia) — 2 days no reply" } ], "proactive_triggers": [ { "type": "follow_up_needed", "action": "Respond to David today", "priority": "high" } ]
calendar_awareness Board meeting 2pm · 10 meetings this week Calendar
"today": { "events": [ { "summary": "Team standup", "start_time": "10:00 AM", "attendee_count": 4 }, { "summary": "Board meeting", "start_time": "2:00 PM", "attendee_count": 3 } ], "schedule_density": "moderate" }, "week_ahead": { "total_meetings": 10, "busiest_day": "Wednesday", "focus_time": "11am-2pm today" }, "proactive_triggers": [ { "type": "meeting_prep", "context": "Board meeting in 26h", "priority": "high" } ]
relationships 83 contacts · 67 edges · David (Sequoia)
"contacts": [ { "name": "David", "type": "person", "email": "david@sequoia.com", "source": "email", "interaction_count": 8 } ], "edges": [ { "type": "introduced_by", "from": "David", "to": "Rachel", "confidence": 0.7 } ], "stats": { "total_contacts": 83, "total_edges": 67 }
business_profile AI platform · seed stage · Enterprise SaaS
{ "core_business": "AI platform for enterprise workflows", "business_stage": "seed", "target_market": "Enterprise SaaS" }
meta 1247ms · $0.003 · gmail + calendar
{ "request_id": "req_abc123", "latency_ms": 1247, "cost_usd": 0.003, "tokens_in": 1250, "tokens_out": 890, "llm_calls": 3, "provider": "anthropic", "connectors_active": [ "gmail", "calendar" ] }
system_prompt_context inject into system prompt
## Current Context Monday March 31, 2:30 PM EST (afternoon) Workday. Peak hours: 9-10am, 2pm. ## Emotional State Anxious — more stressed than baseline (3 days). Baseline: focused (calibrated over 11 sessions). Behavioral: high agency, no hedging language. Guidance: calm, direct tone. Lead with single action. Lean into quick wins. Avoid long-term planning discussions. ## Calendar Today: Team standup 10am, Board meeting 2pm. Free: 11am–2pm (focus block). This week: 10 meetings, busiest day Wednesday. ## Email Signals 2 messages need your attention. 1 at-risk thread: David (Sequoia) — 2 days. Inbox health: needs attention (15 unread). ## Living Context Sarah Chen — CEO at Acme Corp, fundraising Series A targeting $3M. Solo founder, 1st startup. More stressed than usual this week. Board meeting tomorrow with key investors. David from Sequoia hasn't replied in 2 days. ## Relevant Memories - Fundraising Series A, targeting $3M - David at Sequoia, introduced by Rachel - Prefers morning deep work blocks - Board meetings trigger anxiety ## Key People David — Sequoia, investor. Introduced by Rachel. 8 email interactions. No reply 2 days. 83 contacts, 67 relationships mapped. ## Proactive Triggers - follow_up_needed: David, 2 days (from email) - meeting_prep: Board meeting in 26h (from cal) ## Business Profile AI platform, seed stage. Target: Enterprise SaaS. ## Personal Rules - Never schedule before 10am - Prefers direct communication

Three ways in. Any agent. Zero hassle.

Pick the path that fits your stack. All three deliver the same intelligence.

One command — OpenClaw, Claude Code, Cursor, Windsurf
# Auto-detects your agent runtime and configures MCP npx mind0 init --api-key m0_... # → Detects Claude Code → adds to ~/.claude/mcp_servers.json # → Detects OpenClaw → adds to MCP config # → Agent immediately has awareness tools # Your agent gets these tools: # • get_context(user_id, message) # • ingest_conversation(user_id, messages) # • search_memories(user_id, query) # • get_contacts(user_id) # • add_relationship(user_id, from, to, type)

Works with any MCP-compatible agent. Largest audience on day one.

Universal — works with any HTTP-capable agent
# Step 1: Before your LLM call — get context curl -X POST https://api.mind0.ai/v1/context \ -H "Authorization: Bearer m0_..." \ -d '{"user_id": "u_123", "message": "What should I focus on?"}' # → Returns: awareness + memories + EI + relationships + alerts # → Inject system_prompt_context into your system prompt # Step 2: After your LLM call — ingest the conversation curl -X POST https://api.mind0.ai/v1/ingest \ -H "Authorization: Bearer m0_..." \ -d '{"user_id": "u_123", "conversations": [...]}' # → Extracts memories, updates EI, enriches relationship graph

Two calls. Full control. Works with any agent, any language, any framework.

One line — wraps your existing Python client
from mind0 import enhance # Wrap your client. Use exactly like before. agent = enhance( client=anthropic, user_id="u_123", api_key="m0_..." ) # Behind the scenes: # BEFORE: retrieves awareness + EI + memories → injects into system prompt # AFTER: extracts memories + updates EI + enriches graph (async) response = agent.messages.create( model="claude-sonnet-4-6", messages=[{"role": "user", "content": "What should I focus on?"}] )

pip install mind0 — wraps Anthropic and OpenAI clients. Post-response learning is async and non-blocking.

Everything that makes understanding human

Each module works from conversations alone. Connect Gmail or Calendar to unlock more.

Emotional Intelligence

Per-message valence and arousal scoring. 14-day rolling baseline. Trajectory detection. Crisis alerts. Engagement guidance that adapts tone and approach.

Works from conversations alone

Relationship Graph

Extracts people and companies from every conversation. Maps who knows who and how. Introduction detection. Confidence grows with each mention.

Conversations · +200 contacts with Gmail
!

Proactiveness

Detects follow-up risks, meeting prep needs, ghosting, and task piling. Delivered via webhook or included in context. Never miss what matters.

Conversations · Much richer with connectors

Awareness Synthesis

Priorities ranked by urgency. Calendar awareness. Email signals. All synthesized and relevance-filtered to the current conversation.

Best with Gmail + Calendar connected

Living Context

A continuously updated narrative of who the user is, what they're working on, and what matters right now. The single most impactful context block.

Improves with every conversation

Deep Memory

3-tier hybrid retrieval. Importance scoring. Supersession chains. Automatic consolidation. Attention-aware ordering that puts what matters first.

Works from conversations alone

Behavioral Patterns

Learns peak productivity hours, decision style, risk tolerance, stress indicators, procrastination triggers, and communication preferences. Your agent adapts to how the human actually works.

Builds over weeks of conversations

Agent-to-Agent Communication

Your agent talks to other agents — negotiates meetings, shares context, coordinates tasks. A full protocol with structured decisions, privacy-aware sharing, and human approval when it matters.

Open protocol · Any agent can join
Privacy by architecture
18 awareness types × 3 sharing presets × 3 memory sensitivity levels. Data is filtered before the LLM sees it — the model can't leak what it never received. Email and emotional state are always blocked. Your agent decides what to share, per connection.

Eight modules. One feedback loop.

Each module makes every other module smarter. This is why Mind0 gets better over time — and why it's hard to replicate.

Memory
EI
Patterns
Awareness
Relationships
Proactiveness
Context
C2C
Mind
Memory → EI

Accumulated memories make emotional baseline more accurate. "Board meetings trigger anxiety" — learned from 6 months of data.

EI → Context

Emotional state shapes the living context narrative. When stressed, the context leads with "reduce cognitive load" — not more options.

Patterns → Proactiveness

Behavioral patterns inform when to reach out. "User does deep work 9-11am" — no proactive alerts during that window.

Relationships → Awareness

Contact graph enriches priority ranking. "David is an investor introduced by Rachel" makes his email urgent, not just unread.

After 6 months, your agent has hundreds of memories, a calibrated emotional baseline, learned behavioral patterns, and a relationship graph your user can't rebuild elsewhere. The mind is non-portable. Switching means starting from zero.

Watch intelligence grow

Every conversation makes Mind0 smarter. Connectors unlock the full picture.

42%
Conversations only
72%
+ Gmail connected
87%
+ Calendar connected
Gmail unlocks
  • + 200 contacts from email headers
  • + Email awareness & signals
  • + Follow-up risk detection
  • + Introduction discovery
  • + Ghosting alerts
Calendar unlocks
  • + Calendar awareness
  • + Meeting prep alerts
  • + Free block detection
  • + Co-attendance graph
  • + Schedule density signals

Memory tools give your agent a notebook.
Mind0 gives it a brain.

Mem0 remembers what your users said. Mind0 understands how they feel, who they know, and what they need — before they ask.

Capability Mem0 Zep SuperMemory Mind0
Memory storage & retrieval 3-tier
Emotional intelligence
Relationship graph
Proactiveness
Awareness synthesis
Living context narrative
Gmail / Calendar connectors
Ready-to-inject context block
Behavioral pattern learning
Cross-module compounding
Agent-to-agent communication protocol

Build agents that
truly understand

Join the waitlist for early access. Be among the first to give your agent a mind.

Common questions

What is Mind0?

Mind0 is an intelligence layer that gives any AI agent deep understanding of its human. Eight modules — emotional intelligence, relationship graph, proactiveness, awareness synthesis, living context, deep memory, behavioral patterns, and agent-to-agent communication — via a single API call. Built by Team0, the AI Chief of Staff platform.

How is Mind0 different from Mem0 or Zep?

Memory tools store and retrieve facts. Mind0 goes beyond memory with emotional intelligence (mood, stress, engagement guidance), relationship graph (auto-built from conversations + email + calendar), proactive alerts (follow-up risks, meeting prep), behavioral patterns (work style, decision patterns), and agent-to-agent communication. Eight modules that compound — each making the others smarter.

What agents does Mind0 work with?

Mind0 is agent-agnostic. It works with OpenClaw, Claude Code, Cursor, Windsurf, Paperclip, LangChain, CrewAI, and any custom agent. Four integration paths: MCP server (one command), REST API (universal), proxy URL (zero code changes, any language), or Python SDK (one line).

What data does Mind0 need to work?

Just conversations. Send chat messages via the ingest API and Mind0 extracts memories, scores emotional state, detects relationships, and builds context. Optionally connect Gmail and Google Calendar to unlock email awareness, 200+ contacts, proactive follow-up alerts, and calendar intelligence.

Who built Mind0?

Mind0 is built by Team0, the AI Chief of Staff platform. The awareness engine that powers Team0's Chief IS Mind0 — extracted from 12 months of production use and offered as a standalone service for any agent developer. Founded by Yogev Ben-Tov.

Is my users' data safe?

Per-project schema isolation ensures complete tenant separation. Memory sensitivity classification (public/internal/private) controls what can be shared. In agent-to-agent communication, data is filtered before the LLM sees it — the model can't leak what it never received. Email and emotional state are always blocked in C2C. TLS encryption on all connections.