Product Documentation

MarketiQ AI is an AI marketing team in the cloud, built by Trayarunya Ventures. 40 AI agents run one loop — research → strategy → creation → approval → publishing → measurement — across 30+ connected modules for B2B and B2C go-to-market. Every external effect (publishing, sending, spending) is gated by your approval policy; the agents are software, not human staff, and the scope of “marketing” they cover is a maintained capability catalog, not everything.

The product family includes:

• Web platform (marketiqgpt.com) — the dashboard with all modules.
• Chrome extension — LinkedIn profile analysis, connection-note drafts, explicit current-page lead capture and observed-activity sync.
• LinkedIn Copilot desktop app (macOS) — a research and drafting cockpit with isolated sessions and review queues.
• Client portal — white-label approvals, reports and attribution for your clients.
• Developer platform — outbound webhooks, scoped API keys and an MCP server (see section 6).

• Founder / marketing lead: sign up at /signup, complete onboarding (brand, audience, goals), connect channels under Settings → Integrations, run Deep Research on your website, then review the first Master Strategy draft. Nothing ships until you approve it.
• Agency operator: create a workspace per client (Agency plan: 3 included), invite the client to their portal, and set the approval policy and exception owner per client before enabling any loop.
• Developer: under Settings → Developer platform, create a webhook endpoint (store the signing secret once), issue a scoped API key, or register an MCP client and copy the generated command into your MCP host.
• Compliance / DPO: read Data handling, GDPR and Security; each shows its published version, review status and effective date, and the GDPR page hosts the rights-request form.

• Deep Research: market, competitor and audience research plus watchtower and reputation monitoring, based on the public sources and connected data the workspace can actually read.
• Master Strategy: an AI strategist turns evidence into a proposed plan — positioning, pillars, calendar, campaigns and budget pacing — for your review.
• Creation Studio: on-brand posts, articles, ad creatives, video and decks drafted from your Content DNA and brand guardrails.
• Approve & publish: drafts wait in the approval queue; publishing and ad execution run only through catalogued native adapters, or are labelled as client-assisted where no certified adapter exists.
• Measure & learn: analytics, attribution, experiments and modelling report what connected providers return. Modelled outputs are labelled as modelled; measured outcomes are never invented.

• Pipeline: lead pipeline, ABM, ICP builder, forms and referral tracking.
• Brand & Growth: brand kit, Content DNA, guardrails, content studio, decks, calendar and publishing.
• Intelligence: research, insights, watchtower, reputation, experiments, forecast, MMM, attribution and CRO.
• B2B Engine: LinkedIn outreach drafting, influencer discovery and the AI prospect coach (via the Chrome extension). Final LinkedIn actions are performed by you in LinkedIn's UI.
• Automation: AI team (agents), tasks, chat and budget pacing — each loop is approval-gated by default.
• Account: team roles, white-label portals, integrations, model settings, billing and the developer platform.

Connections use each provider's official API over OAuth — we never ask for your social passwords. Whether an operation is native, permitted browser automation, client-assisted, unavailable or prohibited is defined per entry in the capability catalog shown above, not by this prose.

• Social & publishing: LinkedIn, X/Twitter, Meta (Facebook & Instagram), YouTube. TikTok Direct Post is native only for approved apps/accounts with video.publish; otherwise the hand-off is labelled assisted.
• Analytics & search: Google Analytics, Google Search Console.
• Ads: Google Ads (including Ad Grants), Meta, LinkedIn and X execution. TikTok, Microsoft Advertising and Amazon Ads connectors are compatibility-staged and unavailable until explicitly activated.
• CRM & commerce: HubSpot, Salesforce, Microsoft Dynamics, Pipedrive, Shopify and WooCommerce.
• Email data: Klaviyo profile, list-membership, delivery-event and explicit-consent synchronisation.

Connections can be added or revoked at any time from Settings → Integrations; each requests only the minimum scopes for the features you use — see Data handling.

• Outbound webhooks: create endpoints per workspace, choose events from the versioned event catalog, receive HMAC-signed deliveries with a signing secret shown once, rotate secrets, pause/resume, inspect delivery receipts and redeliver failed deliveries.
• Public API keys: keys are scoped (for example research:read, leads:write), optionally rate-limited per minute, expire on a date you set, and rotate with a grace window. Every public API response carries an api_version.
• MCP server: register an MCP client, obtain a client secret once, exchange it at the token endpoint and connect your MCP host. The tool list you see is the negotiated set for your client's scopes; rate budgets are enforced per client and reported on 429.

All of this lives in Dashboard → Settings → Developer platform (owner/admin roles).

• Roles: owner, admin, manager, editor and viewer control what each member can see and do. Invite teammates from Settings → Team.
• Approvals & exceptions: the client (or the workspace owner) owns approval policy and exception handling; agents propose, humans decide on anything with external effect.
• White-label client portals: agencies give each client a branded portal with approvals, reports and attribution without exposing the main workspace.

Plans: Starter $425/mo, Growth $749/mo, Agency $1,249/mo, Enterprise from $5,000/mo; annual billing saves 25%. Manage subscription, payment details and invoices from Settings → Billing. Refunds follow the refund policy and are shown with their real provider status.

Open a case from the Support page (you receive a case id and confirmation state) or your private support inbox. Include the workspace, request/run id, expected result and redacted error. A submitted case is not an acknowledged or resolved case; inspect its actual status and replies.

• 401/403: sign in, verify your address and check workspace membership, role, scoped key and connected-provider grant. Do not retry with someone else's credentials.
• 402/entitlement unavailable: inspect Billing. A paid checkout is not activation until verified provider reconciliation and account verification complete. Missing authoritative terms block work; support must repair them.
• 429: respect Retry-After and lower concurrency. Do not rotate identities to bypass a limit.
• Provider timeout / unknown outcome: inspect the original receipt before retrying. A timeout is not proof of failure. Staff may reconcile an exact provider object; a new request key must not duplicate an uncertain send or charge.
• Blocked approval / stale draft: inspect the current artifact version and policy, regenerate a preview, then obtain a fresh authorized approval. Earlier approvals do not authorize a changed artifact.

• Security — controls, disclosure channel and the vulnerability report form.
• Privacy Policy, Terms of Service, Refund Policy — each rendered from its published, reviewed version when one exists.
• GDPR — rights-request form and jurisdictions served.
• Data handling — subprocessors, consent bases and retention.

• Research → content: use your verified account in a workspace with available entitlement and a small approved budget. Set the brand/source scope, preview research and drafts, review citations and rights, and approve the exact current version. Inspect the run and output; queued work is not completion. Reconnect or correct the rejected input before an eligible retry.
• Publishing → measurement: connect a supported test account and verify operation-specific scopes. Preview the destination and current content, approve through the workspace queue, then inspect the provider id and receipt. Read metrics only for connected, supported sources; zero, missing and modelled values are different. Already dispatched actions may require manual/provider reconciliation after a pause.
• Migration: back up the source, use the migration console to dry-run a supported CSV mapping, review warnings and consent, then explicitly execute. Inspect counts and the recorded snapshot before rollback. Unsupported objects remain in the original system; never assume a full vendor-stack migration.
• Billing / refunds: use Billing to inspect contracted terms, invoices and a nonbinding change preview. Only provider confirmation settles billing state. Refund requests require billing review and a distinct approver; provider uncertainty stays unresolved. Unsupported credit conversion is not presented as a balance adjustment.
• Privacy: verify a rights request through the rights intake. Export delivery uses an expiring encrypted artifact; shared organization erasure requires an ownership review. Deletion verification and retention exceptions are separate facts. Feedback and hosted capture default to a 365-day engineering notice, not a legal finding.

Use a dedicated test workspace and revoke test grants/keys afterward. Provider setup, human review, legal review and a staffed support roster are prerequisites that this documentation does not create.