AImarkOS Docs

Platform Overview

AImarkOS is an AI-Powered Marketing Operating System that replaces fragmented tool stacks with a unified intelligence layer. The platform is built on four core modules that share data, learn together, and execute your marketing across every channel.

Quick Start Guide

Get your AImarkOS workspace operational in under 30 minutes by following these steps:

Connecting Integrations

AImarkOS connects to your existing marketing stack. All integrations are configured in Settings → Integrations.

Ad Platforms

• Google Ads — campaigns, keywords, search terms, conversion data
• Meta (Facebook/Instagram) Ads — ad sets, creatives, audience performance
• LinkedIn Ads — B2B targeting and lead gen campaign data
• TikTok Ads — video performance and conversion tracking

CRM & Email

• HubSpot — contacts, deals, lifecycle stages, email engagement
• Salesforce — lead and opportunity data, pipeline signals
• Mailchimp / Klaviyo — email list sync and campaign performance
• Amazon SES — transactional and campaign email sending (shared pool on Free/Starter; BYO domain on Agency)

Analytics

• Google Analytics 4 — traffic, conversion funnels, audience segments
• Segment — unified event stream and customer profiles
• Mixpanel — product analytics and user behavior signals

Communications

• Gmail / Outlook — email inbox unification
• Twitter/X, Instagram, LinkedIn — social DM inbox
• Intercom / Zendesk — support ticket routing
• Twilio — SMS channel (Agency plan)

Plans & Features

CrewAI Orchestration

The CrewAI module lets you deploy coordinated teams of specialized AI agents that collaborate to complete complex marketing tasks autonomously. Each agent has a defined role, goal set, and tool access — working in concert like a real marketing team.

Agent Roles

AImarkOS ships with the following built-in agent archetypes. You can customize goals, constraints, and tool access for each:

StrategyAgent

Analyzes campaign objectives, market data, and historical performance to produce prioritized briefs, channel recommendations, and budget allocation proposals. Outputs a structured brief consumed by downstream agents.

ContentAgent

Generates copy, subject lines, ad headlines, scripts, and long-form content using your brand voice profile from Content Studio. Supports multi-variant output for Bandit testing.

AnalystAgent

Ingests performance signals from connected integrations, identifies trends and anomalies, and produces plain-language insight summaries. Feeds findings back to the StrategyAgent loop.

PublisherAgent

Executes approved content and campaigns to connected ad platforms and email providers. Requires human-in-the-loop approval by default. Can be set to fully autonomous on Pro and Agency plans.

InboxAgent

Monitors the Unified Inbox, triages incoming messages by priority and intent, and drafts responses using brand voice. Escalates to human team members based on configured rules.

Tasks & Crew Configuration

A Crew is a named group of agents assigned to a campaign goal. To configure a crew:

1. Navigate to CrewAI → New Crew
2. Define your campaign objective in natural language
3. Select agent roles and set their individual goals
4. Configure task dependencies (sequential or parallel execution)
5. Set human-approval gates at key checkpoints
6. Assign tool access per agent (ad platform APIs, CRM, etc.)
7. Click Deploy Crew

Example Crew Config# Example crew definition (YAML format)
crew:
  name: "Q3 Campaign Launch"
  goal: "Generate and launch a full-funnel Q3 campaign targeting SMB decision-makers"
  agents:
    - role: StrategyAgent
      goal: "Produce prioritized channel brief for Q3"
      tools: [analytics_read, crm_read]
    - role: ContentAgent
      goal: "Generate 3 email variants and 8 ad headlines"
      tools: [content_studio, brand_voice]
    - role: PublisherAgent
      goal: "Schedule approved content to Google Ads and email"
      tools: [google_ads_write, ses_send]
      approval_required: true
  execution: sequential

RL Bandit

The RL Bandit module applies multi-armed bandit algorithms to continuously test and reallocate budget across campaign variants — channels, creatives, or audience segments — based on live performance signals. Unlike traditional A/B testing, bandits explore and exploit simultaneously, minimizing wasted spend.

Configuring Arms

An arm represents a variant you want to test. Arms can be:

• Channel arms — e.g., Email vs. Paid Search vs. Social
• Creative arms — e.g., Headline A vs. Headline B vs. Headline C
• Audience arms — e.g., Segment 18-34 vs. Segment 35-54
• Offer arms — e.g., 10% discount vs. Free trial vs. Demo

To create a Bandit campaign:

1. Go to RL Bandit → New Campaign
2. Name your campaign and define your arms (minimum 2, no maximum on Pro/Agency)
3. Set your reward signal: conversion, roas, ctr, or custom metric
4. Configure initial budget allocation (equal split recommended)
5. Select your algorithm (see below)
6. Set minimum exploration budget per arm (prevents premature elimination)
7. Launch — the Bandit begins reallocating within the first 24 hours

Algorithms

UCB (Upper Confidence Bound) — Recommended for new campaigns

UCB balances exploration and exploitation by selecting arms with the highest upper confidence bound on expected reward. Performs best when you have limited prior data and want to minimize regret early. Recommended for most campaigns.

Thompson Sampling — Recommended for high-volume campaigns

Thompson Sampling uses Bayesian inference to model reward distributions per arm, sampling from each distribution to select the arm to run next. Performs best with high impression volume and provides more nuanced exploration. Recommended for campaigns with 10,000+ daily impressions.

Epsilon-Greedy — Simple baseline

Exploits the current best arm (1-ε)% of the time and explores randomly ε% of the time. Simple and interpretable, but less efficient than UCB or Thompson. Useful for debugging or low-stakes tests.

Content Studio

Content Studio generates on-brand marketing content at scale — governed by your brand voice and automatically formatted for each channel and use case. All generated content can be routed directly into RL Bandit campaigns for performance testing.

Brand Voice Setup

Brand Voice is the foundation of Content Studio. Before generating content, set up your voice profile:

1. Go to Content Studio → Brand Voice
2. Upload 5–20 examples of your best-performing content (emails, ads, blog posts, social copy)
3. Define tone attributes: choose from a matrix of dimensions (formal↔casual, authoritative↔friendly, concise↔expansive)
4. Add brand vocabulary: words and phrases to always use or always avoid
5. Set topic boundaries: subjects the AI should not engage with on your behalf
6. Click Train Voice Profile — processing takes 2–5 minutes

You can maintain separate voice profiles per sub-account (Pro and Agency plans).

Output Formats

Email Campaigns

Generate subject lines, preview text, and full email body copy. Specify audience segment, campaign objective, and desired CTA. Output supports plain text and HTML templates.

Paid Ads

Generate Google responsive search ad headlines and descriptions, Meta ad copy (primary text, headline, description), and LinkedIn Sponsored Content. Multi-variant output auto-routes to RL Bandit arms.

Social Posts

Generate native-formatted posts for LinkedIn, Twitter/X, Instagram, and Facebook. Includes character count enforcement and hashtag suggestions per platform.

Long-Form Content

Generate blog posts, case study outlines, whitepaper sections, and landing page copy. SEO optimization layer suggests keyword placement and meta descriptions.

Image Generation

Available on Pro and Agency plans. Generate on-brand visual assets for ads and social using text-to-image AI. Specify brand colors, style guidelines, and aspect ratio.

Unified Inbox

Unified Inbox consolidates every customer conversation — across email, social DMs, live chat, and support — into a single AI-powered interface. The InboxAgent triages, prioritizes, and drafts responses using your brand voice, so your team can focus on high-value conversations.

Connecting Channels

To connect a channel, navigate to Inbox → Channels → Add Channel:

• Email — connect Gmail or Outlook via OAuth; or configure a custom IMAP/SMTP mailbox
• Twitter/X DMs — connect via Twitter API credentials
• Instagram DMs — connect via Meta Business API
• LinkedIn Messages — connect via LinkedIn OAuth
• Live Chat — embed the AImarkOS chat widget on your site, or connect Intercom/Zendesk
• SMS — connect Twilio (Agency plan only)

Free plan supports email + 2 social channels. Starter/Pro support all channels except SMS. Agency adds SMS and custom webhook channels.

Routing & Triage Rules

Configure how the InboxAgent handles incoming messages:

Priority Scoring

The AI assigns a priority score (High / Medium / Low) based on: sender type (existing customer, lead, unknown), message sentiment, intent classification (complaint, inquiry, sales, support), and keyword triggers you define.

Auto-Assignment

Route conversations to specific team members or queues based on priority, channel, keyword, or sender segment. Configure in Inbox → Settings → Routing Rules.

Auto-Draft

The InboxAgent drafts a response to every incoming message using your brand voice. Drafts appear in the composer and require one-click approval before sending. You can configure fully autonomous sending for specific message types (e.g., FAQ responses) in Pro and Agency plans.

Managing Sub-Accounts

Sub-accounts allow you to manage multiple client workspaces from a single agency login. Each sub-account has its own data, campaigns, brand voice, and inbox — fully isolated from other sub-accounts.

Creating a Sub-Account

1. Go to Agency → Sub-Accounts → New Account
2. Enter client name, contact email, and business details
3. Assign the sub-account to a team member (optional)
4. Set up integrations and brand voice for the client workspace

Billing Management

Agency plan accounts can centralize billing for all sub-accounts. Your clients are billed through your agency account — you control what they see and what markup, if any, you apply.

Sub-Account Limits by Plan

• Free — not available
• Starter — 1 sub-account
• Pro — up to 5 sub-accounts
• Agency — unlimited sub-accounts

White-Label Setup

Agency plan subscribers can white-label AImarkOS under their own brand. This includes:

• Custom domain — serve the platform from app.youragency.com
• Custom logo and brand colors in the UI
• Custom email sender domain — send platform notifications from your domain
• Removal of all AImarkOS branding from the client-facing interface
• Custom login page and onboarding flows

Domain Setup

1. Go to Agency → White-Label → Custom Domain
2. Enter your subdomain (e.g., app.youragency.com)
3. Add the provided CNAME record to your DNS provider
4. DNS propagation typically takes 24–48 hours
5. SSL is provisioned automatically via Let's Encrypt

Frequently Asked Questions

Does AImarkOS replace my ad platform accounts?

No. AImarkOS connects to your existing Google Ads, Meta Ads, and other platform accounts via API. You retain full ownership of your ad accounts and campaigns. AImarkOS manages and optimizes within those accounts.

How does the RL Bandit know what's working?

The Bandit receives reward signals from your connected conversion tracking — Google Ads conversions, Meta pixel events, CRM deal creation, or custom webhook events. It uses these signals to update arm weights and reallocate budget toward higher-performing variants.

Is AI-generated content safe to publish without review?

We recommend reviewing all AI-generated content before publishing. The platform includes a review queue for all AI outputs. You can configure auto-publish rules for low-risk content types (e.g., social posts) while requiring manual approval for ads and emails.

What LLMs power the AI features?

AImarkOS uses a mix of OpenAI (GPT-4o) and Anthropic (Claude) models depending on the task. You can specify a preferred provider in Settings → AI Preferences. Enterprise plans support custom model endpoints.

Can I export my data?

Yes. All campaign data, analytics, and content can be exported in CSV or JSON format from Settings → Data Export. Pro and Agency plans additionally support automated PDF report generation.

Is my client data kept separate from other users?

Yes. Each account and sub-account operates in an isolated data environment. Client data is never shared across accounts or used to train shared models without explicit consent.

What happens if I exceed my plan limits?

You'll receive an in-app notification when approaching limits. The platform will not automatically upgrade or charge you for overages — you'll be prompted to upgrade your plan. Free accounts exceeding 3 campaigns will have new campaigns paused until an older one is archived or deleted.

Support & Contact

Our team is here to help. Choose the right channel for your needs: