Agent Planners

Overview

Agent Planners is an AI agent for marketing and growth work. You describe a goal. An orchestrator and six specialist agents build a plan, run it across your connected ad, analytics, commerce, and publisher platforms, and report back. Reads run freely, but by default every write waits for your approval and every change is logged. It suits performance marketers, ad-ops teams, agencies that manage many client accounts, and publishers.

Key Features

Multi-Agent System: An orchestrator hands work to six specialist agents covering analytics, optimization, creative, research, safety, and reporting. You can choose the model each one uses.
Approval Gates and Audit Log: Campaign, budget, bid, and keyword changes wait for your OK in the app or through a one-time email link. Every change is logged with who made it, when, the channel, and before/after details.
Stale-State Checks: The agent re-checks live account data before it runs a write. Plans built on outdated numbers are rejected instead of applied.
29 Platform Connections: It connects natively to Google Ads, GA4, Search Console, BigQuery, DV360, CM360, Meta, TikTok, Microsoft Advertising, LinkedIn, Shopify, Stripe, HubSpot, Klaviyo, and more. You can reach 250+ more apps through Composio, plus your own custom MCP servers.
Reusable Workflows and Scheduling: The agent turns your goals into readable plans you can re-run or schedule with cron in any timezone. Each scheduled run re-checks live data.
Agent Harness and Memory: Built-in tools include web search, Google Trends, Firecrawl, Exa, a hosted browser, image and video generation, and a Python sandbox. The agent also remembers account details like targets, protected campaigns, and excluded geos.
Shareable Dashboards:Create dashboards from templates or ask the agent to build one. Share dashboards through public links that you can revoke. You can also generate reports and client-ready slide decks.
MCP, REST API, and Agent Mail: You can use Agent Planners from Claude, Codex, Cursor, or ChatGPT over MCP, or call it through the REST API. You can also email tasks in, and nothing runs until you confirm.

Pros

Human Control by Default: You approve writes before they run. Deleting campaigns, moving money, and publishing always need a person, even in hands-off mode, which makes it safer to point at live ad accounts.
One Workflow Across Many Platforms: You can plan and review work across ads, analytics, and your store in one place instead of switching between dashboards.
Marketing Decisions Based on Store Data: Connect Shopify, GA4, Tag Manager, and advertising platforms to compare campaign performance with store activity and revenue data.
Flexible Plans: A free entry point lets you try the platform before upgrading. Check plan limits for credits, task concurrency, and advanced features.
Bring Your Own Model: On paid plans, you can point agents at your own OpenAI-compatible endpoint. LLM tokens on that endpoint don't use platform credits.
Agency-Friendly Workspace Management: Keep client accounts, integrations, audit logs, and billing separate across organizations, with roles for different team members.

Cons

Limited Free Usage: The Free plan's 2,500 monthly credits cover about one task a month. That's enough to try the product, but not to run real optimization work on it.
Key Features Need a Paid Plan: Bring Your Own Model and the MCP server's tools mode require a paid subscription. Buying top-ups alone does not unlock them.

Use Cases

Google Ads Optimization: You can ask the agent to review account performance and suggest negative keywords and bid or budget changes. You approve each change before it goes live, and the audit log records exactly what changed.
E-commerce Growth: You can connect Shopify, GA4, Tag Manager, Google Ads, and Meta Ads and have the agent tie campaign performance to store results. Then you can act on what is actually driving sales.
Agency Client Management: Run separate organizations and workspaces for each client, with role-based access and API keys limited to specific platforms. Scheduled runs and dashboards keep each account monitored without manual check-ins.
Recurring Reporting: Set up a scheduled dashboard from one of seven templates and share it with stakeholders through a link you can revoke. Each refresh pulls current data, so nobody waits for a manual report.
Publisher Revenue Monitoring: Connect Google Ad Manager, AdSense, and AdMob to track monetization. Email alerts flag metric thresholds or account status problems.
Marketing Ops from Your AI Assistant: If you already work in Claude, Cursor, Codex, or ChatGPT, you can call Agent Planners over MCP. You can run account analysis and queue approved changes without leaving that tool.

Frequently Asked Questions

What is Agent Planners?

Agent Planners is a multi-agent AI platform for marketing and growth work. An orchestrator and six specialist agents plan, run, and review tasks across your connected ad, analytics, commerce, and publisher platforms. By default, changes need your approval.

Can the agent change my ad accounts without my permission?

Not by default. The agent can read data freely, but every write waits for your approval in the app or through a one-time email link. Deleting campaigns, ad groups, or ads, moving money, and publishing always need a person, even in hands-off mode.

How much does Agent Planners cost?

Agent Planners has a Free plan with 2,500 monthly credits and no credit card required. Paid plans are Lite at $29.90/month, Starter at $100/month, Scale at $500/month, and Agency at $1,000/month. One-time credit top-ups are also available, starting at 10,000 credits for $15. Subscription credits reset each billing period, while top-up credits do not expire. Check the official pricing page for current annual billing discounts.

Which platforms does it integrate with?

Agent Planners has 27 native integrations and 2 built-in data sources. The native list includes Google Ads, GA4, Search Console, BigQuery, DV360, CM360, Meta, TikTok, LinkedIn, Microsoft Advertising, Shopify, Stripe, HubSpot, and Klaviyo, and DataForSEO and Context.dev are built in. You can reach 250+ more apps through Composio and add your own MCP servers.

Can I use my own AI model?

Yes, on paid plans. You can point agents at any OpenAI-compatible endpoint, and LLM tokens used on it don't consume platform credits. A top-up alone doesn't unlock this feature.

Can I use Agent Planners from Claude or ChatGPT?

Yes. It works over MCP with hosts such as Claude, Codex, Cursor, and ChatGPT, and it also offers a REST API. The MCP server's tools mode requires a paid subscription.

Is it suitable for agencies?

Yes. It supports multiple organizations with admin, editor, and member roles, isolated workspaces, and API keys limited to specific platforms. There's also a dedicated Agency plan with 1,200,000 monthly credits and up to 30 tasks running at once.