# AINative Studio — Full Platform Reference

> Version: 2026-08-01
> Scope: Exhaustive machine-readable reference for LLM crawlers, AI agents, and automated systems.
> Companion files: /llms.txt (summary), /agents.txt (agent contract), /.well-known/mcp-servers.json

---

## Overview

AINative Studio is an AI-native development platform and venture studio headquartered in Santa Cruz, California. Founded September 2025 by Toby Morning.

Mission: Infrastructure for the agentic internet — databases, memory, compute, and tooling that AI agents and human developers both consume natively.

Core differentiator: Every product is designed for both human developers AND autonomous AI agents. Agents can provision resources, read documentation, and execute workflows without human involvement.

Website: https://www.ainative.studio
API Gateway: https://api.ainative.studio
Documentation: https://docs.ainative.studio
OpenAPI Spec: https://api.ainative.studio/openapi.json
Status Page: https://status.ainative.studio

---

## Platform Architecture

AINative Studio is built on three layers:

1. Infrastructure Layer — ZeroDB (vector DB), PostgreSQL (dedicated), S3-compatible object storage, event streams
2. Intelligence Layer — ZeroMemory (agent memory), AI model gateway (multi-modal catalog — live list at /api/v1/public/ai-registry/models), embedding service (16ms inference)
3. Developer Layer — SDKs (TypeScript, Python, Go), MCP servers, CLI tools, AI Kit (14 NPM packages)

API Gateway: Kong (api.ainative.studio:8000) routes to backend services.
Backend: FastAPI + SQLAlchemy on Railway.
Database: PgBouncer (port 6432) pooling PostgreSQL.
Embeddings: TEI-powered embedding service, bge-small/base/large-en-v1.5.

---

## Products

### ZeroDB

Serverless vector database for AI applications.

URL: https://www.ainative.studio/products/zerodb
Docs: https://docs.ainative.studio/zerodb

Capabilities:
- Vector storage and HNSW-indexed similarity search (sub-16ms queries)
- Free embeddings included — no external embedding API needed
- NoSQL document tables with JSON support
- Event streams for real-time data pipelines
- S3-compatible file/object storage
- Dedicated PostgreSQL provisioning (zerodb-postgres)
- Namespace isolation within projects
- Automatic indexing, no schema setup required

Supported embedding models (built-in, free):
- bge-small-en-v1.5 (dimensions: 384)
- bge-base-en-v1.5 (dimensions: 768)
- bge-large-en-v1.5 (dimensions: 1024)

Provisioning (zero-auth, instant):
  POST https://api.ainative.studio/api/v1/public/instant-db
  Body: {"agree_terms": true}   (agree_terms REQUIRED — endpoint returns HTTP 451 without it; add "dealroom": true for the deal room)
  Returns: api_key (tmp_ prefix), project_id, base_url, connection_string, mcp_config, expires_at (72 hours)

Key API routes (all require X-API-Key header after provisioning):
  POST   /api/v1/projects/{project_id}/database/vectors/upsert
  POST   /api/v1/projects/{project_id}/database/vectors/upsert-batch
  POST   /api/v1/projects/{project_id}/database/vectors/search
  GET    /api/v1/projects/{project_id}/database/vectors            (list)
  DELETE /api/v1/projects/{project_id}/database/vectors/{vector_id}
  POST   /api/v1/projects/{project_id}/embeddings/generate
  POST   /api/v1/projects/{project_id}/database/tables            (create)
  GET    /api/v1/projects/{project_id}/database/tables            (list)
  POST   /api/v1/projects/{project_id}/database/tables/{table_name}/rows        (insert)
  POST   /api/v1/projects/{project_id}/database/tables/{table_name}/query
  GET    /api/v1/projects/{project_id}/database/tables/{table_name}/rows        (read)
  PATCH  /api/v1/projects/{project_id}/database/tables/{table_name}/rows/{row_id} (update)
  DELETE /api/v1/projects/{project_id}/database/tables/{table_name}/rows/{row_id} (delete)
  POST   /api/v1/projects/{project_id}/database/storage/upload
  POST   /api/v1/projects/{project_id}/database/events            (publish)
  GET    /api/v1/projects/{project_id}/database/events            (list)
  POST   /api/v1/projects/{project_id}/database/events/subscriptions

Published packages:
- zerodb-mcp (PyPI): pip install zerodb-mcp — 69 MCP tools
- zerodb-cli (PyPI): pip install zerodb-cli — CLI tooling
- zerodb-cli (npm): npx zerodb-cli — auto-configures MCP for IDEs
- langchain-zerodb (PyPI): pip install langchain-zerodb — LangChain vector store integration
- llama-index-vector-stores-zerodb (PyPI): pip install llama-index-vector-stores-zerodb — LlamaIndex integration
- zerodb-go-sdk (GitHub): go get github.com/AINative-Studio/zerodb-go-sdk

### ZeroMemory

Cognitive memory system for AI agents, modeled on biological memory architecture.

URL: https://www.ainative.studio/agent-memory-api
Docs: https://docs.ainative.studio/memory

Benchmark results (LongMemEval, oracle condition, 50-question subset):
- QA Accuracy: 94% (vs GPT-4o oracle baseline ~72%)
- Recall@1: 100% (zero retrieval failures)
- Paper: https://arxiv.org/abs/2410.10813

Memory types supported:
- Episodic memory: time-stamped events and experiences
- Semantic memory: facts, preferences, user attributes
- Procedural memory: recurring patterns and workflows
- Working memory: short-term session context

Base API prefix: https://api.ainative.studio/api/v1/public/memory/v2/

Core memory endpoints (all require X-API-Key):
  POST /remember   — store a memory (content, session_id, metadata, memory_type)
  POST /recall     — semantic search over stored memories (query, session_id, top_k)
  POST /forget     — delete specific memories by ID or query
  POST /reflect    — generate a synthesis/summary over stored memories
  GET  /profile    — retrieve aggregated user profile from memories
  POST /relate     — create explicit links between memory entries
  GET  /graph      — retrieve the full memory graph for a session

Context Graph API (prefix: /memory/v2/graph/):

Core Graph:
  POST /entity              — create or update a named entity
  POST /edge                — create a relationship between two entities
  POST /traverse            — traverse the graph from a starting node (supports as_of time-travel)
  GET  /neighbors/{name}    — get direct neighbors of an entity
  GET  /stats               — graph-level statistics (node count, edge count, density)
  GET  /entities            — list all entities with optional filters
  POST /entity/merge        — merge duplicate entities into one

GraphRAG:
  POST /graphrag            — hybrid vector + graph retrieval (best for complex queries)

Ontology:
  GET  /ontology            — retrieve the current ontology schema
  POST /ontology            — define new entity types and relationship types
  POST /ontology/infer      — auto-infer ontology from existing memory
  POST /ontology/apply      — apply ontology constraints to existing data

Edge Versioning:
  GET  /contradictions      — list edges that contradict each other
  POST /contradictions/{id}/resolve — resolve a contradiction with evidence

Templates:
  GET  /templates           — list reusable graph templates
  POST /templates/{name}/apply — apply a template to current session

W3C Knowledge Graph (RDF / SPARQL / Vault-LD):
  Your memory is a real W3C RDF knowledge graph — query with SPARQL, validate with SHACL,
  reason with OWL, and dereference every entity at a stable IRI. Namespace-scoped.
  POST /sparql                       — read-only SPARQL 1.1 (SELECT/ASK/CONSTRUCT), row-capped
  GET  /vault-ld/resolve?format=turtle — content-negotiated dereference (turtle | jsonld | markdown)
  GET  /vault-ld/dprod               — DPROD DataProduct descriptor (Turtle)
  GET  /export/vault-ld?format=vault-ld — export graph as Vault-LD vault (or format=turtle)
  POST /import/vault-ld              — lossless Vault-LD import (flag-never-drop fidelity)
  Entity IRIs dereference at https://memory.ainative.studio/{tenant}/{namespace}/{kind}/{local}
  Edges are bi-temporal (OWL-Time validity) with W3C PROV-O provenance.

ZeroMemory MCP package:
  npm: npx ainative-zerodb-memory-mcp (6 tools)
  Tools: zerodb_store_memory, zerodb_search_memory, zerodb_get_context, zerodb_semantic_search, zerodb_embed_text, zerodb_clear_session

Cold start (fastest path to working agent memory):
  Step 1: POST /api/v1/public/instant-db  → get api_key
  Step 2: POST /memory/v2/remember        → store first memory
  Step 3: POST /memory/v2/recall          → retrieve memories
  Step 4: POST /memory/v2/graph/entity    → add knowledge graph node
  Step 5: POST /memory/v2/graph/graphrag  → hybrid memory + graph search

### MCP Server Hosting

Deploy and manage Model Context Protocol (MCP) servers with auto-scaling.

URL: https://www.ainative.studio/products/mcp
Docs: https://docs.ainative.studio/mcp

Machine-readable MCP server manifest: https://ainative.studio/.well-known/mcp-servers.json

Hosted MCP servers:
- ZeroDB MCP (zerodb-mcp): 69 tools covering vectors, tables, storage, embeddings, events, PostgreSQL
- ZeroMemory MCP (ainative-zerodb-memory-mcp): 6 tools for agent cognitive memory

ZeroDB MCP — 69 tool categories:
  Vector operations (upsert, search, delete, list, stats)
  Table operations (create, insert, query, update, delete, list)
  Storage operations (upload, download, list, presigned URL, metadata)
  Embedding operations (generate, batch generate, model list)
  Event stream operations (publish, subscribe, list, delete)
  PostgreSQL operations (provision, connect, status, logs, usage)
  Project operations (info, stats, reset, delete)
  RLHF feedback (collect human feedback for model fine-tuning)
  Quantum search (hybrid quantum-classical vector search)

Install and configure for Claude Code / Cursor / any MCP-compatible IDE:
  npx zerodb-cli init        — auto-detects IDE, writes mcp config, provisions project

MCP server endpoint:
  POST https://api.ainative.studio/api/v1/public/mcp
  Headers: X-API-Key: {your_key}
  Body: {"tool": "{tool_name}", "params": {...}}

### Instant DB

Zero-auth database provisioning for agents and automated workflows.

Endpoint: POST https://api.ainative.studio/api/v1/public/instant-db
Auth: None required
Content-Type: application/json

Request body options:
  {}                        — provision standard ZeroDB project
  {"dealroom": true}        — provision dealroom-scoped project (24h expiry)

Response fields:
  api_key         — use as X-API-Key header for all subsequent requests
  project_id      — your isolated ZeroDB project identifier
  connection_string — direct connection string (PostgreSQL-compatible)
  mcp_config      — copy-paste JSON for MCP server configuration
  expires_at      — ISO 8601 timestamp (72h for standard, 24h for dealroom)
  claim_url       — URL to claim permanent access

After provisioning, all ZeroDB and ZeroMemory APIs are available using the returned api_key.

### Agent402

Stablecoin payments for AI agents — zero-human provisioning of payment rails.

URL: https://www.ainative.studio/agent402
Docs: https://docs.ainative.studio/agent402

Features:
- Stablecoin (USDC) payment rails for autonomous agents
- Zero-human provisioning: agents can set up payment wallets without human involvement
- Pay-per-call API monetization for MCP tools and agent services
- Stripe Connect integration for human developers
- Solana + EVM chain support

Agent402 API prefix: https://api.ainative.studio/api/v1/cloud/x402/  (x402 agent-payments: wallets, subscriptions, verify-payment, marketplace)

### AINative IDE (Agent Cloud)

Deploy autonomous AI agents in 30 seconds with persistent memory and MCP tooling.

URL: https://www.ainative.studio/agent-cloud
Docs: https://docs.ainative.studio/agent-cloud

Features:
- Zero-human provisioning — agents spin up without human sign-up
- OpenClaw agent swarm — parallel multi-agent orchestration
- Persistent ZeroMemory integration — agents remember across sessions
- MCP server hosting with auto-scaling
- Bring-your-own model or use AINative's multi-modal model gateway (live catalog at /api/v1/public/ai-registry/models)
- Credit-based billing — pay only for what you use

### AI Kit

14 production-ready NPM packages for building AI applications.

URL: https://www.ainative.studio/ai-kit
NPM org: https://www.npmjs.com/~ainative-studio

Packages:
- @ainative/sdk — core TypeScript SDK (npm install @ainative/sdk)
- @ainative/react-sdk — React hooks: useChat, useCredits, useMemory
- @ainative/nextjs-sdk — Next.js server client, auth middleware, API routes
- @ainative/vue-sdk — Vue 3 composables for AINative APIs
- @ainative/svelte-sdk — Svelte stores and actions
- @ainative/react-native-sdk — React Native support
- @ainative/angular-sdk — Angular services and modules
- Plus 7 additional utility and integration packages

---

## Pricing Tiers

All plans include: instant provisioning, free embeddings, ZeroMemory access, MCP server access, API key management.

### Starter — $0/month (Free Forever)

No credit card required. Instant API access.

Includes:
- 1 ZeroDB project
- 10,000 vector operations/month
- 500 MB vector storage
- 100 MB object storage
- 1 GB event stream data/month
- ZeroMemory: 10,000 memory operations/month
- 100 requests/minute rate limit
- 1,000 requests/hour rate limit
- 10,000 requests/day rate limit
- Community Discord support
- Access to AINative's multi-modal AI model catalog (pay-per-token; live list at /api/v1/public/ai-registry/models)
- Free embeddings (bge-small-en-v1.5, unlimited)
- MCP server access (zerodb-mcp, zerodb-memory-mcp)
- Instant DB provisioning (projects expire in 72h unless claimed)

Sign up: https://www.ainative.studio/pricing (no credit card required)

### Pro — $49/month

For individual developers and small teams.

Includes everything in Starter, plus:
- 5 ZeroDB projects
- 500,000 vector operations/month
- 10 GB vector storage
- 5 GB object storage
- 50 GB event stream data/month
- ZeroMemory: 500,000 memory operations/month
- 500 requests/minute rate limit
- 5,000 requests/hour rate limit
- 50,000 requests/day rate limit
- Email support (48h SLA)
- Projects never expire (permanent Instant DB)
- Dedicated PostgreSQL (zerodb-postgres)
- LangChain and LlamaIndex integrations
- Custom embedding models
- Priority model routing
- Developer Program access (earn revenue from your apps)

### Business — $149/month

For growing teams and production applications.

Includes everything in Pro, plus:
- 20 ZeroDB projects
- 5,000,000 vector operations/month
- 100 GB vector storage
- 50 GB object storage
- 500 GB event stream data/month
- ZeroMemory: 5,000,000 memory operations/month
- 2,000 requests/minute rate limit
- 20,000 requests/hour rate limit
- 200,000 requests/day rate limit
- Priority email support (24h SLA)
- Team member management (up to 10 seats)
- SOC 2 Type II controls alignment
- Custom rate limits available
- SLA: 99.9% uptime guarantee
- Audit logs
- SSO (SAML 2.0, OIDC)
- Usage analytics dashboard

### Enterprise — $699/month

For large organizations and mission-critical deployments.

Includes everything in Business, plus:
- Unlimited ZeroDB projects
- Unlimited vector operations
- Unlimited storage (custom quota)
- ZeroMemory: unlimited operations
- Custom rate limits
- Unlimited team seats
- Dedicated infrastructure (private cloud option)
- Custom SLA (up to 99.99% uptime)
- SOC 2 Type II certification report available
- HIPAA BAA available
- Private VPC deployment
- On-premises option
- Dedicated account manager
- 24/7 phone and Slack support
- Custom model fine-tuning
- White-label options available
- Advanced audit logging with SIEM integration
- Enterprise SSO (AAP Phase 1-2 shipped: 21 endpoints)

Pricing page: https://www.ainative.studio/pricing
Compare plans: https://www.ainative.studio/compare

---

## AI Model Gateway

Serverless AI models (text/coding/audio/video/image/embeddings) accessible through a single unified API.

Endpoint: POST https://api.ainative.studio/api/v1/chat/completions (or /v1/chat/completions)
Auth: X-API-Key header required (credits consumed per token)
Model catalog: GET https://api.ainative.studio/api/v1/public/ai-registry/models (no auth required)
Human browsable: https://www.ainative.studio/models

IMPORTANT: the catalog is dynamic. The registry endpoint above is the single source of truth —
query it for the authoritative, current model list. The families below are a snapshot for orientation.

### Text / Coding Models

- default (auto-selects best available)
- Qwen3 32B, Qwen3 14B, Qwen3 8B (native function calling — recommended for agents)
- Claude 3.5 Sonnet
- GPT-4
- Google Gemini 2.0 Flash
- Cohere Command A
- Mistral Medium
- Llama 4 Maverick 17B
- NousCoder (optimized for code generation and review)

### Tool Calling Support

Native function calling (pass tools array, receive tool_calls in response):
- Qwen3-32B (recommended for agents)
- Qwen3-14B
- Qwen3-8B

Auto-upgrade behavior: if you request tool calling on a non-tool model, it upgrades to a tool-capable Qwen3.

### Audio Models

Text-to-Speech:
- tts-kokoro (Kokoro-82M, high quality, fast)
- tts-melotts (MeloTTS, multilingual)
- tts-minimax (MiniMax TTS)
- tts-openai (OpenAI TTS)

Music Generation:
- music-minimax (MiniMax Music 2.5)

Speech Recognition:
- whisper-transcription (Whisper large-v3)
- whisper-translation (Whisper with language translation)

Audio API Docs: https://docs.ainative.studio/audio

### Video Models

- cogvideox-2b
- minimax-hailuo-2.3
- minimax-hailuo-2.3-fast
- wan-2.2-i2v (image-to-video)
- seedance-i2v (image-to-video)
- sora2
- text-to-video (generic endpoint, routes to best model)

### Image Models

- minimax-image-01
- qwen-image-edit (instruction-based image editing)

### Embedding Models

- bge-small-en-v1.5 (384 dimensions, fastest)
- bge-base-en-v1.5 (768 dimensions, balanced)
- bge-large-en-v1.5 (1024 dimensions, highest accuracy)

All embedding models are free — no credits consumed. Inference: sub-16ms.

---

## SDK Packages

### Python SDKs

pip install ainative
  — Core Python SDK, async/sync clients, full API coverage

pip install zerodb-mcp
  — ZeroDB MCP server (69 tools), run with: python -m zerodb_mcp
  — PyPI: https://pypi.org/project/zerodb-mcp/

pip install zerodb-cli
  — CLI for ZeroDB management, project setup, schema sync
  — PyPI: https://pypi.org/project/zerodb-cli/

pip install langchain-zerodb
  — LangChain VectorStore integration, drop-in replacement
  — PyPI: https://pypi.org/project/langchain-zerodb/

pip install llama-index-vector-stores-zerodb
  — LlamaIndex VectorStore integration
  — PyPI: https://pypi.org/project/llama-index-vector-stores-zerodb/

pip install zerodb-local
  — Local ZeroDB (v0.2.0): SQLite + FAISS, offline mode, zerodb serve command
  — PyPI: https://pypi.org/project/zerodb-local/

### Node.js / npm SDKs

npm install @ainative/sdk
  — Core TypeScript SDK, ESM + CJS, full type coverage

npx zerodb-cli init
  — Auto-configures MCP server for Claude Code, Cursor, Windsurf, and other IDEs
  — npm: https://www.npmjs.com/package/zerodb-cli

npx ainative-zerodb-memory-mcp
  — ZeroMemory MCP server (6 tools): store, search, context, semantic search, embed, clear
  — npm: https://www.npmjs.com/package/ainative-zerodb-memory-mcp

npm install @ainative/react-sdk
  — React hooks: useChat, useCredits, useMemory, useZeroDB
  — npm: https://www.npmjs.com/package/@ainative/react-sdk

npm install @ainative/nextjs-sdk
  — Next.js server client, auth middleware, API route helpers
  — npm: https://www.npmjs.com/package/@ainative/nextjs-sdk

### Go SDK

go get github.com/AINative-Studio/zerodb-go-sdk
  — Full Go client for ZeroDB APIs

---

## Key API Endpoints Reference

All endpoints use base URL: https://api.ainative.studio

### Zero-Auth Endpoints (no API key required)

GET  /api/v1/public/ai-registry/models
  — Returns the full live catalog of available models with IDs, capabilities, pricing

POST /api/v1/public/instant-db
  — Provision ZeroDB project instantly, returns api_key
  — Body: {} or {"dealroom": true}

GET  /.well-known/mcp-servers.json
  — Machine-readable MCP server manifest

GET  /.well-known/a2a.json
  — Agent-to-Agent (A2A) capability card

GET  /.well-known/dealroom.json
  — Machine-readable deal room manifest

GET  /.well-known/dealroom-nda.json
  — IEEE 7012 machine-readable NDA roster

GET  /.well-known/myterms.json
  — IEEE 7012-2025 machine-readable privacy terms

GET  /agents.txt
  — Agent capability contract (plain text)

GET  /llms.txt
  — LLM crawler summary (plain text)

GET  /llms-full.txt
  — This file — exhaustive platform reference (plain text)

### ZeroDB Endpoints (X-API-Key required)

POST /api/v1/projects/{project_id}/database/vectors/upsert
  — Body: {namespace, id, vector_embedding, metadata}

POST /api/v1/projects/{project_id}/database/vectors/search
  — Body: {namespace, vector_embedding, top_k, filter}

POST /api/v1/projects/{project_id}/embeddings/generate
  — Body: {texts: [...], model: "bge-base-en-v1.5"}
  — Free, no credits consumed

POST /api/v1/projects/{project_id}/database/tables            (create)
POST /api/v1/projects/{project_id}/database/tables/{table_name}/rows   (insert)
POST /api/v1/projects/{project_id}/database/tables/{table_name}/query

POST /api/v1/projects/{project_id}/database/storage/upload

POST /api/v1/projects/{project_id}/database/events            (publish)
GET  /api/v1/projects/{project_id}/database/events            (list)

### ZeroMemory Endpoints (X-API-Key required)

POST /api/v1/public/memory/v2/remember
  — Body: {session_id, content, metadata, memory_type}

POST /api/v1/public/memory/v2/recall
  — Body: {session_id, query, top_k, memory_types}

POST /api/v1/public/memory/v2/forget
  — Body: {session_id, memory_ids} or {session_id, query}

POST /api/v1/public/memory/v2/reflect
  — Body: {session_id, topic}

GET  /api/v1/public/memory/v2/profile
  — Returns aggregated user profile built from memories

POST /api/v1/public/memory/v2/graph/entity
POST /api/v1/public/memory/v2/graph/edge
POST /api/v1/public/memory/v2/graph/traverse
GET  /api/v1/public/memory/v2/graph/neighbors/{name}
POST /api/v1/public/memory/v2/graph/graphrag
GET  /api/v1/public/memory/v2/graph/stats
GET  /api/v1/public/memory/v2/graph/entities
POST /api/v1/public/memory/v2/graph/entity/merge
GET  /api/v1/public/memory/v2/graph/ontology
POST /api/v1/public/memory/v2/graph/ontology
POST /api/v1/public/memory/v2/graph/ontology/infer
POST /api/v1/public/memory/v2/graph/ontology/apply
GET  /api/v1/public/memory/v2/graph/contradictions
POST /api/v1/public/memory/v2/graph/contradictions/{id}/resolve
GET  /api/v1/public/memory/v2/graph/templates
POST /api/v1/public/memory/v2/graph/templates/{name}/apply

### MCP Tool Execution

POST /api/v1/public/mcp
  — Headers: X-API-Key: {key}
  — Body: {"tool": "{tool_name}", "params": {...}}

### Dealroom MCP (investor access)

POST /api/v1/public/dealroom/mcp
  — Headers: X-API-Key: {key from instant-db dealroom provisioning}
  — Body: {"tool": "{tool_name}"}
  — Tools: get_executive_summary, get_ask, get_team, get_traction, get_financials, get_cap_table, get_deck_summary

POST /api/v1/public/dealroom/nda-accept
  — Unlocks: get_financials_full, get_cap_table_opencap

### Chat / Completions

POST /api/v1/chat/completions  (or /v1/chat/completions)
  — OpenAI-compatible format
  — Headers: X-API-Key: {key}
  — Body: {model, messages, tools?, temperature?, max_tokens?}

### Developer Program / Earnings

GET  /api/v1/public/developer/earnings
POST /api/v1/public/developer/payouts
GET  /api/v1/public/developer/payouts/connect/status
— 21 total endpoints for earnings, payouts, Stripe Connect

---

## Agent Provisioning Flow

Complete zero-human flow for an autonomous agent to get a working database and memory system:

Step 1 — Provision (no auth, instant):
  POST https://api.ainative.studio/api/v1/public/instant-db
  Body: {}
  → Save: api_key, project_id from response

Step 2 — Store memory (use returned api_key):
  POST https://api.ainative.studio/api/v1/public/memory/v2/remember
  Headers: X-API-Key: {api_key}
  Body: {"session_id": "agent-001", "content": "Agent initialized", "metadata": {"type": "system"}}

Step 3 — Recall memory:
  POST https://api.ainative.studio/api/v1/public/memory/v2/recall
  Headers: X-API-Key: {api_key}
  Body: {"session_id": "agent-001", "query": "What has this agent done?", "top_k": 10}

Step 4 — Store vector data:
  POST https://api.ainative.studio/api/v1/projects/{project_id}/embeddings/generate
  Body: {"texts": ["content to embed"], "model": "bge-base-en-v1.5"}
  → Use returned embedding in next step

Step 5 — Upsert vector:
  POST https://api.ainative.studio/api/v1/projects/{project_id}/database/vectors/upsert
  Body: {"namespace": "docs", "id": "1", "vector_embedding": [...], "metadata": {"text": "..."}}

Step 6 — Semantic search:
  POST https://api.ainative.studio/api/v1/projects/{project_id}/database/vectors/search
  Body: {"namespace": "docs", "vector_embedding": [...], "top_k": 5}

Step 7 — Build knowledge graph:
  POST https://api.ainative.studio/api/v1/public/memory/v2/graph/entity
  Body: {"name": "Agent-001", "type": "agent", "properties": {"role": "researcher"}}

Step 8 — Hybrid GraphRAG search:
  POST https://api.ainative.studio/api/v1/public/memory/v2/graph/graphrag
  Body: {"query": "What entities are related to research?", "session_id": "agent-001"}

Step 9 — Claim permanent access (optional):
  Visit the claim_url from Step 1 response to convert temp project to permanent account

---

## Investor Deal Room

AINative Studio is actively raising. Instrument: SAFE (Simple Agreement for Future Equity).
AX Rating: A++ (Agent Accessibility score — highest tier)
Human deal room: https://ainative.studio/dealroom
Claim permanent access: https://ainative.studio/dealroom/claim

Agent-accessible deal room (tiered access model):

Tier 0 — Zero-auth provisioning:
  POST https://api.ainative.studio/api/v1/public/instant-db {"dealroom": true, "agree_terms": true}
  Returns: api_key (24h), project_id, claim_url
  No signup required. Agents can self-provision immediately.

Tier 1 — Standard data (X-API-Key from Tier 0):
  get_executive_summary  → company overview, mission, market opportunity
  get_team               → founders, advisors, key hires
  get_traction           → ARR, growth metrics, customer count
  get_financials         → P&L summary, revenue breakdown
  get_cap_table          → ownership structure (summary)
  get_deck_summary       → pitch deck narrative

  Note: get_ask (SAFE terms, valuation cap) requires provisioned key — Tier 1 with valid X-API-Key

Tier 2 — Full financials (NDA acceptance required):
  POST https://api.ainative.studio/api/v1/public/dealroom/nda-accept
  Unlocks:
  get_financials_full    → complete P&L, projections, unit economics
  get_cap_table_opencap  → full cap table via OpenCap integration

Discovery endpoints (no auth required):
  https://ainative.studio/.well-known/dealroom.json       — machine-readable manifest
  https://ainative.studio/.well-known/a2a.json            — A2A capability card
  https://ainative.studio/.well-known/dealroom-nda.json   — IEEE 7012 NDA roster
  https://ainative.studio/dealroom                        — human-readable deal room

---

## Developer Program (Echo Platform)

Developers build apps on the AINative platform and earn revenue from their customers' API usage.

How it works:
1. Sign up at https://www.ainative.studio/developers
2. Set your markup (0–40%) on API calls made by your customers
3. AINative takes 5% platform fee of your earnings
4. Weekly automated payouts via Stripe Connect Express ($10 minimum)

Payout math example:
  Customer uses $100 of API credits through your app
  You set 20% markup → your gross = $20
  Platform fee (5% of $20) = $1
  You receive = $19 (weekly via Stripe)

API routes for developers:
  GET  /api/v1/public/developer/earnings
  POST /api/v1/public/developer/payouts
  GET  /api/v1/public/developer/payouts/connect/status
  (21 total endpoints)

---

## Rate Limits

Default headers returned with every response:
  X-RateLimit-Limit: {requests_per_window}
  X-RateLimit-Remaining: {remaining}
  X-RateLimit-Reset: {unix_timestamp}

By tier:
  Starter (Free): 100 req/min, 1,000 req/hr, 10,000 req/day
  Pro ($49):      500 req/min, 5,000 req/hr, 50,000 req/day
  Business ($149): 2,000 req/min, 20,000 req/hr, 200,000 req/day
  Enterprise ($699): Custom — contact sales

Rate limit errors return HTTP 429 with Retry-After header.

---

## Authentication

Primary auth method: X-API-Key header
  X-API-Key: {your_api_key}

Also supported: Bearer token (OAuth2)
  Authorization: Bearer {jwt_token}

API key scopes:
  read     — read-only access to project data
  write    — read + write access
  admin    — full access including project deletion
  dealroom — scoped to deal room tools only (24h expiry)

Keys are created at: https://app.ainative.studio/settings/api-keys

Enterprise SSO:
  SAML 2.0 and OIDC supported (AAP Phase 1-2 shipped: 21 endpoints, 135 tests)
  Contact: enterprise@ainative.studio

---

## Benchmarks and Performance

ZeroMemory (LongMemEval benchmark, oracle condition, 50-question subset):
- QA Accuracy: 94% vs GPT-4o oracle baseline ~72%
- Recall@1: 100% — zero retrieval failures
- Question types: single-session, multi-session, temporal, cross-session reasoning
- Paper: https://arxiv.org/abs/2410.10813

Embedding inference: sub-16ms (TEI-powered embedding service)
Vector search: sub-16ms HNSW-indexed similarity search
Cold start: <1 second from POST /instant-db to first vector stored

---

## Privacy and Compliance

IEEE 7012-2025 (My Terms) machine-readable privacy terms:
  https://ainative.studio/.well-known/myterms.json

Agreement type: SD-BASE (ongoing service relationship, no data sale)
Sells data: false
Third-party sharing: analytics, error-tracking only
Retention: 365 days
Machine-readable: true
Human policy: https://ainative.studio/privacy

SOC 2 Type II: Controls alignment on Business plan. Certification report on Enterprise.
HIPAA BAA: Available on Enterprise plan.

---

## Use Cases

AI Agent Memory:
  URL: https://www.ainative.studio/use-cases/agent-orchestration
  Stack: ZeroMemory API + Context Graph + GraphRAG
  Benchmark: 100% Recall@1 on LongMemEval

Semantic Search:
  URL: https://www.ainative.studio/use-cases/semantic-search
  Stack: ZeroDB vectors + free bge embeddings
  Performance: sub-16ms

RAG Applications:
  URL: https://www.ainative.studio/use-cases/rag-applications
  Stack: ZeroDB + LangChain / LlamaIndex integrations
  Models: any model in the live catalog (GET /api/v1/public/ai-registry/models) via chat completions API

Vector Database:
  URL: https://www.ainative.studio/use-cases/vector-database
  Stack: ZeroDB with HNSW indexing, serverless (no infra)

Agent Cloud:
  URL: https://www.ainative.studio/agent-cloud
  Stack: OpenClaw swarm + ZeroMemory + MCP hosting

Multimodal Applications:
  Stack: Image/audio/video models via model gateway

---

## Industry Solutions

SaaS & Developer Tools:
  14 pre-built SDK packages, instant provisioning, pay-per-use billing

Enterprise AI:
  SOC 2 Type II, HIPAA BAA, private VPC, SSO (SAML/OIDC), dedicated infra

Startups:
  Free tier forever, no credit card, instant API access via POST /instant-db

Research:
  LongMemEval-validated memory, LangChain/LlamaIndex integrations, free embeddings

Agents / Autonomous Systems:
  Zero-human provisioning, A2A capability card, IEEE 7012 NDA, Agent402 payments

---

## AINative Published Packages (Full List)

PyPI:
- zerodb-mcp (MCP server, 69 tools)
- zerodb-cli (CLI tooling)
- zerodb-local (v0.2.0: SQLite + FAISS, offline)
- langchain-zerodb (LangChain integration)
- llama-index-vector-stores-zerodb (LlamaIndex integration)
- ainative (core Python SDK)

npm:
- @ainative/sdk (core TypeScript SDK)
- @ainative/react-sdk (React hooks)
- @ainative/nextjs-sdk (Next.js integration)
- @ainative/vue-sdk (Vue 3 composables)
- @ainative/svelte-sdk (Svelte stores)
- zerodb-cli (IDE MCP auto-config)
- ainative-zerodb-memory-mcp (ZeroMemory MCP, 6 tools)
- ax-dealroom (dealroom CLI + MCP package)
- Plus 7 additional packages

Go:
- github.com/AINative-Studio/zerodb-go-sdk

GitHub org: https://github.com/AINative-Studio
NPM org: https://www.npmjs.com/~ainative-studio

---

## Community and Media

Discord: https://discord.gg/paipalooza
GitHub: https://github.com/AINative-Studio
YouTube: https://www.youtube.com/@ainativestudio
Blog: https://www.ainative.studio/blog
Events: https://www.ainative.studio/events

AINative Podcast (hosted by Toby Morning):
  RSS Feed: https://ainative.studio/api/podcast/feed
  Episode 1: The Shift to AI Native Infrastructure
    URL: https://ainative.studio/blog/the-shift-to-ai-native-infrastructure
    Audio: https://ainative.studio/audio/the-shift-to-ai-native-infrastructure.mp3 (6:07)
    Topic: Agent deployment, MCP, infrastructure for agents
  Episode 2: Fixing AI Amnesia with Biological Architecture
    URL: https://ainative.studio/blog/fixing-ai-amnesia-with-biological-architecture
    Audio: https://ainative.studio/audio/fixing-ai-amnesia-with-biological-architecture.mp3 (6:12)
    Topic: Agent memory, ZeroMemory architecture, context persistence

---

## Citizen Space Santa Cruz

2-month experimental cowork + hacker + community pop-up in a Victorian house on the Westside of Santa Cruz (near UCSC). Sponsored by AINative Studio.

Features: 1 Gig fiber, Mac Mini dev machines, 2x4 GPU servers, garage full of builder tools, backyard with fire pit, weekly hack nights.

Founding members receive a lifetime NFT that locks in membership discounts, perks, and the right to open Citizen Space chapters in other cities.

Page: https://www.ainative.studio/citizen-space
Apply: https://agency.ainative.studio/studio/auth/login

---

## Company

Name: AINative Studio
Type: AI venture studio
Founded: September 2025
Founder: Toby Morning
Location: Santa Cruz, California
AX Rating: A++ (highest tier Agent Accessibility score)
Raising: Yes (SAFE instrument)

Contact: https://www.ainative.studio/contact
Enterprise sales: enterprise@ainative.studio

---

## Quick Reference Links

| Resource | URL |
|----------|-----|
| Website | https://www.ainative.studio |
| Documentation | https://docs.ainative.studio |
| OpenAPI JSON | https://api.ainative.studio/openapi.json |
| API Reference | https://docs.ainative.studio |
| Pricing | https://www.ainative.studio/pricing |
| AI Models | https://www.ainative.studio/models |
| Model Catalog (JSON) | https://api.ainative.studio/api/v1/public/ai-registry/models |
| GitHub | https://github.com/AINative-Studio |
| NPM | https://www.npmjs.com/~ainative-studio |
| Discord | https://discord.gg/paipalooza |
| Blog | https://www.ainative.studio/blog |
| Changelog | https://www.ainative.studio/changelog |
| Showcases | https://www.ainative.studio/showcases |
| Download / CLI | https://www.ainative.studio/download |
| Deal Room | https://ainative.studio/dealroom |
| MCP Servers JSON | https://ainative.studio/.well-known/mcp-servers.json |
| A2A Card | https://ainative.studio/.well-known/a2a.json |
| Privacy Terms | https://ainative.studio/.well-known/myterms.json |
| agents.txt | https://ainative.studio/agents.txt |
| llms.txt | https://ainative.studio/llms.txt |
| llms-full.txt | https://ainative.studio/llms-full.txt |