Your agents. Your rules.

Build agents.
Give them voice.
Let them learn.

Create agents with distinct personalities, voices, and specialties. They share a memory pool but think for themselves. Calibrate them, adjust them, let them grow. Browser, IDE, phone — they’re wherever you are.

Start building I already have an account

$ madcat agents
    CORE    opus-4   direct, technical     mem: shared + 847 own
    SENSEI  sonnet-4 teaching, patient     mem: shared + 312 own
    CADENCE sonnet-4 audio, creative       mem: shared + 1.2k own
    RAVEN   sonnet-4 github, ops           mem: shared + 94 own
    shared memory pool: 5,412 entries
$ _

Start with the account you trust most

Sign in with GitHub, Google, or a one-time email code. Your agents, conversations, and preferences sync across every surface.

Continue with GitHub Continue with Google

Create and calibrate

Spin up new agents in minutes. Set their personality, pick their voice, define what they’re good at. Calibrate with feedback until they feel right. Adjust anytime.

Shared memory, individual minds

All agents draw from one shared memory pool — project context, preferences, decisions. Each agent also builds its own memory from conversations. They know what matters and what’s theirs.

Specialists that teach

Agents that pair-program, review code, walk you through startup ops. Not just answers — structured teaching from specialists you’ve shaped yourself.

Everywhere you work