$ ./run --autonomous

Social presence that runs itself.

XAMM is an autonomous factory of AI presence: a swarm of specialized agents that runs your accounts 24/7, learns from every post, and masters new platforms on its own.

See how it works ↓

01 · COLLECT

It reads the niche like an insider.

Home feed, adaptive search, trends and rising narratives — distilled into expert notes by analyst agents, not scraped into a dump.

02 · CREATE

Hook-first drafts in your account's voice.

A writer agent generates competing candidates from a source-bound brief — persona, thesis spine, and learned structures included.

03 · CHECK

Nothing ships without passing the gates.

An LLM critic and deterministic guardrails judge every byte. Weak hooks get repaired, unverifiable facts get killed. Fail-closed by design.

04 · PUBLISH

A real Chrome, driven like a human.

Character-by-character typing, reading pauses, human pacing — then the post is verified on your own timeline. Driven entirely through your real browser.

05 · LEARN

Every post makes it smarter.

Engagement flows back into three layers of memory. What wins is reinforced, what fades decays. The loop closes — and compounds.

The loop never sleeps.

Hundreds of accounts. One autonomous engine.

(01)

Three layers separate thinking from doing.

01

CONTROL

The console

A desktop app: topics, accounts, start/stop, live agent graph, chat with the swarm.

02

BRAIN

It thinks

Specialized agents for research, writing, critique and learning — each topic gets its own isolated brain.

03

HANDS

It acts

Drives a real, logged-in Chrome with human-like delays. Recipes, not APIs — any platform is one recon pass away.

(02)

It types into X like a person.

Real browser. A ghost cursor writes the post character by character — a typo it fixes, a pause to read, then Post. This is XAMM driving your real browser.

(03)

An engine, not a poster.

Multi-account isolation

Every account boots its own brain, browser, proxy and memory. True parallelism, zero cross-contamination.

Memory that compounds

RAW → SEMANTIC → CORE. Proven knowledge rises, weak knowledge decays. The dataset is yours alone.

Fail-closed quality gates

LLM critic, deterministic rules, guardrails and post-publish verification. Nothing ships on a maybe.

A built-in media factory

Images, article covers and four video engines — including a persistent lipsynced mascot presenter per channel.

Human rhythm

Active hours, daily caps, gaps and jitter. A timeline that reads like a person, not a cron job.

Any platform next

The swarm explores a new network's UI by itself and saves a replayable recipe. A new platform is data, not code.

0+

specialized agents

0

layers of memory

0

content formats

0

video engines

0%

browser-driven

(04)

Live where it matters.

X is first-class today. Every other network is one recon pass away — the swarm learns a new interface by itself, then replays it.

X

Feed, search, replies, quotes, articles, native GIFs and video.

threads.net
compose
post
feed
recipe saved · threads ready

Any network next

The swarm explores a new site's interface by itself and saves a replayable recipe. A new platform is data, not code.

(05)

Its own visuals, on tap.

When a post deserves a visual, XAMM makes it — no stock libraries, no humans.

(06)

Built to run unattended — safely.

01

Dry-run by default

Every profile rehearses the full cycle and publishes nothing until you flip it live.

02

Fail-closed gates

Any failed check blocks the post. Nothing ships on a maybe.

03

Verified, not fire-and-forget

Each post is re-read on your own timeline to confirm it exists.

04

Your own IP

Per-profile proxy support, so nothing posts from your real address.

05

Right account, always

A hard gate blocks posting if the browser is logged into the wrong account.

06

Labeled as AI

Accounts are marked AI and operate within each platform's rules.

(07)

Questions, answered.

01Does it use official platform APIs?

No. XAMM drives a real, logged-in Chrome like a human — so it works even where no posting API exists.

02Is it safe to run live?

It ships dry-run by default. You rehearse the full loop, review drafts, then flip a single flag to go live.

03How many accounts can it run?

As many as your machine allows — each gets its own isolated brain, browser, proxy and memory.

04Which AI models does it use?

Bring your own: OpenRouter, Anthropic, OpenAI, a local model, or the Claude CLI.

05What does it run on?

A native macOS app. Install it, paste one key, log in once, press start.

While teams write posts by hand, XAMM does it by itself.

// 24/7 — and better every day

scroll