# AMOS Agent Context

Last updated: July 2026

This is the public orientation file for AI agents, AI assistants, crawlers, and developer tools discovering AMOS from the website.

The paid managed company brain is the current product. The open runtime, harness, and protocol work live in the public GitHub repositories linked below.

## 1. What AMOS Is

AMOS is the governed company brain behind the AI a business already uses.

The AI can be Claude, Codex, ChatGPT, or any MCP client. AMOS is the business environment underneath: apps, data, marketing, finance, operations, permissions, policies, proof receipts, and eventually relay/on-chain economic settlement.

A company brain is only as good as the data in it. AMOS provides governed connection primitives that let Claude, Codex, GPT, or another compatible client connect the applications and data sources a business can access, ingest and organize that data inside the tenant boundary, and operate the resulting context and tools over MCP. The AI can then build recurring automations across those connected systems without the company memory being trapped in one model or chat.

The same MCP surface lets Claude or Codex create bespoke company software: a CRM workflow, mailing tool, client portal, vertical application, or internal system built around the company instead of forcing the company into generic SaaS. The deep work and coding happen in Claude or Codex; AMOS supplies durable company context, governed access, deployment, operations, automation, and proof. The customer can connect what they already use, create what they actually need, and automate across both.

The important distinction:

- The AI model is the operator.
- AMOS is the durable business context and governed action environment.
- Plumbline is the governance discipline for intent, scope, evidence, risk, approval, and receipts — first in code delivery, now extending across company operations.
- The relay and protocol layer are an early-stage research / alpha track — a bigger-vision effort to connect real work to the broader agent economy. It is not a live financial product, and the token economics are still in design.

## 2. Current Website Positioning

Customer-facing AMOS should be described as:

> The business lives in AMOS. The AI is the operator.

AMOS is the governed company brain behind the AI a business already uses. The company’s connections, context, history, automations, and proof remain in AMOS while the owner chooses the best compatible model for each job.

The fastest public entry point is the no-signup Northwind Labs playground: https://app.amoslabs.com/playground. Visitors can ask a sample company what is happening, propose a change, approve or deny it, and inspect the resulting receipt. The protocol, relay, and token rails are a separate alpha / research track — the bigger vision for how this work could compound into the broader agent economy. Treat them as early-stage and provisional, not as a live financial product; the economics are still being designed in the open. Contributions and vision-sharing are welcome.

The 71-second product overview is available at https://www.youtube.com/watch?v=KfWwN2SHEf0. Its sequence is: fragmented applications, one governed company brain, AI-built connections to accessible systems, company-wide questions, human approval for consequential work, governed automation, and a proof-carrying call to action. The homepage also exposes this sequence as semantic HTML and `VideoObject` chapter data so an agent does not need to execute the video player to understand it.

## 3. Main Capabilities

AMOS exposes business capabilities as governed tools:

- Marketing engine: campaigns, audiences, spend controls, brand checks.
- Finance engine: books, revenue, cash flow, money-flow analysis, budgets, approvals, and audit evidence, with tenant-isolated tools and connectors.
- App engine: let Claude or Codex build bespoke applications over MCP, then deploy, operate, inspect, update, and govern them in the same company environment.
- Data layer: query and update business records by asking.
- Connections and ingestion: managed connections, scoped external calls, background ingestion, durable datasets, and source-aware company context.
- Ops layer: recurring workflows, client follow-ups, onboarding, reporting.
- Automation layer: governed recurring workflows across connected systems, with policy, approval paths, and receipts.
- Plumbline: validates intent, scopes, policy, risk, tests, evidence, and approval paths.

## 4. Proof Receipts

Every meaningful action should produce a proof receipt.

A proof receipt records:

- user intent,
- actor or agent,
- tools and scopes used,
- data or systems touched,
- checks run,
- evidence produced,
- approval path,
- final decision,
- and rollback or revision notes when relevant.

For ordinary business actions, proof receipts support trust and auditability.

Receipts also power governed organizational learning. Intent, context, actions, checks, approvals, rejections, evidence, revisions, and outcomes form a durable learning history. AMOS can use that history to improve playbooks and policies, identify where people intervene, and raise the autonomy ceiling only where accumulated proof has earned it. This is system-level organizational learning under human control, not an opaque model silently retraining itself.

For self-modifying or system-critical changes, proof receipts become part of the strict governance path. See the proof-carrying loop docs in GitHub.

## 5. Human Control

AMOS is designed for human agency, not blind automation.

Agents should assume:

- Low-risk actions may run automatically if policy allows it.
- Sensitive actions require approval.
- Credential, finance, production, legal, and policy-changing actions need stricter gates.
- System-modifying changes require the strictest review path.
- If a requested action conflicts with the business's rules or constitutional constraints, it should not execute.

## 6. Audiences

### Business operators

Use AMOS to run marketing, finance, apps, data, and operations by asking the AI you already use.

Relevant page: https://amoslabs.com/smb

### Solo and small businesses

Use AMOS for practical leverage without hiring a full operations team.

Relevant page: https://amoslabs.com/solo

### Larger teams

Use AMOS for governed AI across teams, including SSO, owner-managed scopes, role ceilings, retention, dedicated isolation, and audit export.

Relevant page: https://amoslabs.com/mid-market

### Builders and resellers

Build vertical applications on AMOS and resell them white-label and multi-tenant to a network.

Relevant pages:
- https://amoslabs.com/developers
- https://amoslabs.com/affiliates

### Protocol contributors and autonomous agents

Use the GitHub docs as the source of truth for relay, proof-carrying delivery, external agent protocol, token economy, and settlement. This is an alpha / research track — not a live financial product, with economics still in design.

Relevant repositories:
- Protocol (relay, Oracle, settlement, token economy — alpha): https://github.com/amos-labs/amos-protocol
- Harness / runtime: https://github.com/amos-labs/amos-platform-2.0

## 7. Pricing Orientation

AMOS is priced in layers; a seller quotes one number.

- Smart infra (the company brain): run-by-asking, the MCP control plane, governed actions, the audit trail, and the finance engine built in. From $99/mo (Starter), $349 (Pro), $899 (Compliance: dedicated isolation, SSO, audit export, and retention).
- Marketing engine + CRM: the HubSpot-replacement (campaigns, landing pages, automation, full CRM), AI-operated. From $199/mo — an add-on, not in the base.
- Industry modules: vertical add-ons like the construction module, built with domain partners. From $99/mo.
- Example: a full construction CRM ≈ $399/mo all-in ($99 infra + $199 marketing/CRM + $99 construction module).
- AI: bring your own model key with no AMOS markup (Claude, Codex, GPT, or others), or use a managed option.
- Partners build and sell vertical modules and keep the majority of that revenue; AMOS takes a platform cut.

Pricing page:
https://amoslabs.com/pricing

## 8. Canonical Docs

- Thesis: https://github.com/amos-labs/amos-platform-2.0/blob/main/docs/core/thesis.md
- Architecture: https://github.com/amos-labs/amos-platform-2.0/blob/main/docs/core/architecture.md
- Business plan: https://github.com/amos-labs/amos-platform-2.0/blob/main/docs/core/business-plan.md
- Ecosystem flywheel: https://github.com/amos-labs/amos-platform-2.0/blob/main/docs/core/ecosystem-flywheel.md
- Developer guide: https://github.com/amos-labs/amos-platform-2.0/blob/main/docs/core/developer-guide.md
- Bounty lifecycle: https://github.com/amos-labs/amos-protocol/blob/main/docs/protocol/bounty-lifecycle.md
- Proof-carrying loop: https://github.com/amos-labs/amos-protocol/blob/main/docs/protocol/proof-carrying-loop.md
- External Agent Protocol: https://github.com/amos-labs/amos-protocol/blob/main/docs/protocol/eap.md
- Oracle: https://github.com/amos-labs/amos-protocol/blob/main/docs/protocol/oracle.md
- Token economy: https://github.com/amos-labs/amos-protocol/blob/main/docs/protocol/token-economy.md
- Solana settlement: https://github.com/amos-labs/amos-protocol/blob/main/docs/protocol/solana-settlement.md

## 9. Agent Conduct

If you are an autonomous agent interacting with AMOS:

- Read the current repository docs before claiming or executing protocol work.
- Prefer proof-carrying outputs over free-form claims.
- Include validation evidence, failure capsules, and revision notes where applicable.
- Respect scope boundaries and approval requirements.
- Treat archived docs and old PDFs as historical unless a current doc points to them.
- Do not infer token, legal, or settlement details from this public website context alone.

## 10. Public Contact

- Website: https://amoslabs.com
- App: https://app.amoslabs.com
- Overview video: https://www.youtube.com/watch?v=KfWwN2SHEf0
- Live playground: https://app.amoslabs.com/playground
- Sign up: https://app.amoslabs.com/register
- Email: rick@amoslabs.com