Antarious Documentation

Antarious AI is an AI-native operating platform for small businesses and commercial go-to-market teams that need to run marketing, sales, CRM, messaging, and money work at speed — without giving up human accountability. At the centre of the platform is Freya: a persistent, memory-enabled AI Manager who coordinates specialist agents across your commercial operations. Freya is not a chatbot. She observes your connected activity, drafts the work that would otherwise consume your day, waits for a human to approve external actions, then executes at machine speed while recording every step.

The platform was built on a simple thesis: most AI tools either automate too little to be transformative, or too much to be trusted. Antarious rejects that trade-off. Every customer-facing or money-moving action — every post published, every reply sent, every outreach sequence launched, every invoice reminder released — passes through an approve-before-execute gate before it leaves your business. Freya does the busywork. Humans stay accountable.

Three commitments, in priority order.

01 · AI-First, Not AI-Added. Antarious was not built by bolting a language model onto an existing SaaS product. The agent architecture, the memory layer, the approval workflows, and the audit trail were designed together so Freya can draft continuously while governance stays consistent across posts, messages, pipeline moves, and money follow-ups.

02 · Human-in-the-Loop. Every external action passes through a human approval gate. You can tighten or widen routing as confidence builds — single approver for routine captions, multi-step review for high-stakes outreach — but the gate itself is structural. Freya prepares; you authorise.

03 · Full Transparency. Every meaningful output shows context, sources where available, and the approval chain that released it. The audit trail is not a forensic add-on; it is how Antarious AI proves who decided what, and when.

If you take only one thing from this guide: Antarious AI is not a productivity gadget that happens to have AI. It is a governance-first automation platform where Freya is the teammate, the approval workflow is the manager, and the audit trail is the record.

Freya is the orchestration layer. Underneath her sit specialist helpers for commercial work — each focused on a narrow job. Content help drafts captions and variants. Messaging help prepares inbox replies. Customer/pipeline help keeps people and deals tidy. Money help prepares invoices and reminders. Freya routes work, assembles the package, and surfaces it to a human approver.

Most small businesses start with a handful of surfaces — Posts, Messages, Customers, Money, and Home — then add depth as the approval cadence becomes familiar. You do not need every specialist on day one.

Context sharing between specialists is fundamental. When a campaign produces interested people, that context informs follow-up drafts. When a deal advances, money drafts can reference the same commercial story. Freya manages the handoff so workflows stay coherent instead of becoming a pile of disconnected tools.

What Antarious AI is not: it is not a chat box you babysit with prompts all day; not a Zapier-style deterministic chain alone; and not an unsupervised auto-poster. Freya's default posture is internal preparation for your team, under your rules, with external execution gated by named humans.

Week one is about building intuition for Freya's drafts — not maximising send volume. Use this checklist as a practical onboarding path for owners and small teams.

• Day 1 — Account and brand baseline. Create your workspace, invite anyone who should approve work, and give Freya the basics: business name, offer, tone, channels you care about, and who owns final say on posts vs money.
• Day 2 — Connect channels carefully. Connect the social and messaging surfaces you already use. Confirm credentials are stored for the integrations Freya needs, and verify that draft-only mode is understood by everyone on the team.
• Day 3 — Review Home as command centre. Treat the Home view as your morning brief. Expect queued drafts, pending approvals, and follow-ups — not a blank dashboard waiting for prompts.
• Day 4 — Approve deliberately. For every draft: approve as-is, edit then approve, request revision, or reject. Do not rush past risk flags. The goal is a clean habit, not speed theatre.
• Day 5 — Pipeline and money hygiene. Let Freya surface interested people, deal reminders, and invoice drafts. Confirm role boundaries so the wrong person cannot release financial follow-ups.
• Day 6–7 — Tighten what still feels noisy. Adjust alert sensitivity, templates, and who approves which action types. Keep external gates on. Widen only where quality has already proven consistent.

When you are ready to work live, open the product at app.antarious.com.

External actions require a designated human to authorise them. Freya prepares the review package — summary, context, proposed output, risk flags, and decision options. Approvers can approve, approve with edits, request revision, or reject. Pending does not mean silent auto-send.

Read the full model on Human Approval.

Role-based control defines who can see data, who can instruct Freya, who can approve which action types, and who can change configuration. Permission self-escalation is not part of the design: users cannot quietly grant themselves authority they do not have.

Details live on Role-Based Control.

The audit trail records what triggered work, what Freya prepared, what the human saw, what they decided, and what executed. It exists so owners, teammates, and reviewers can answer "who authorised this?" with a named person and a timestamp — not a guess.

Explore Audit Trail.

Business data is encrypted in transit and at rest, isolated by customer organisation, and access-controlled by role. Customer data is processed to provide the Services — not to train shared models for other customers. Incident handling and DPA expectations are covered on the security trust page.

Start with Security.

Start narrow. Posts + Messages + Home is enough for many shops and studios in the first fortnight. Add Customers and Money when the approval habit is solid.

Name approvers explicitly. If two co-owners share a business, decide who signs off content versus cash. Ambiguity creates delayed queues, not safety.

Write the brand once. Offer, tone, exclusions ("never promise X"), and examples of good past posts dramatically improve draft quality. Freya cannot invent a brand memory you never provided.

Use structured help when stuck. The Help Centre covers operational stuck-points; the comprehensive FAQ covers deeper product and trust questions; email support handles account-specific issues. For live work, keep the app open while you resolve the blocker.

Larger commercial GTM deployments — agencies, multi-seat sales teams, heavier integration maps — may use a more formal discovery → configuration → validation → go-live path. The governance principles stay the same: Freya prepares, humans approve, audit records retain the chain.