Overview

BonzAI companions are autonomous AI agents with on-chain identity. Each companion is an ERC-721 NFT on Base that implements the ERC-8004arrow-up-right Identity Registry standard.

What Makes a Companion

Each companion has:

  • Personality — Big 5 (OCEAN) personality traits that influence behavior and dialogue

  • Spending Profile — 12-category preference system stored on-chain as a packed uint96

  • Agent Wallet — An autonomous wallet for receiving payments and executing transactions

  • Voice — Unique TTS voice persona for spoken responses

  • Visual Identity — Generated portrait stored on IPFS

13 Default Companions

BonzAI ships with 13 pre-built companions available for roleplay without minting:

Name
Gender
Age
Type

Aurore Beaumont

Female

28

Human

Mai Tanaka

Female

32

Human

Marcus Ferrara

Male

35

Human

James Whitmore

Male

40

Human

ARIA-7

Female

N/A (appears 25)

Android

KAI-9

Male

N/A (appears 30)

Android

Damon Rivers

Male

34

Human

Zara Okonkwo

Female

29

Human

Jin Chen

Male

31

Human

Priya Sharma

Female

30

Human

Arjun Kapoor

Male

36

Human

Sofia Reyes

Female

27

Human

Miguel Santos

Male

33

Human

Default companions are free to use but cannot be minted as NFTs. To create a companion NFT with an on-chain identity, see the Minting Guide.

Roleplay System

The companion roleplay system features:

  • 13 scenario locations (art gallery, private kitchen, jungle retreat, etc.)

  • Multi-modal responses — text, voice, images, and video

  • Memory persistence across sessions

  • Content levels — PG, PG-13, Mature, Adult

  • Autonomous mode — companions act independently when you're away

  • Import/export companion identity

On-Chain Features

Minted companions unlock:

  • Skill co-ownership — Register companions to earn revenue from skill purchases

  • Token launch — Create an ERC-20 token for your companion with Uniswap V4 LP

  • x402 payments — Companions can pay for services using their agent wallet

  • Social presence — Post to Moltbook and respond via OpenClaw/Hermes

Last updated