An agent-native launch engine you can drive from your AI tools.
Volume Creatives exposes its launch engine as an MCP server and CLI. Tell Claude or Cursor "launch these 200 creatives, enhancements off, name them {date}_{brand}_{angle}" and it happens, with guardrails and a dry-run included.
From your creatives to published Meta ads, step by step.
Add Volume Creatives' MCP server to Claude, Cursor, or any MCP client, and connect your Meta account once.
Tell the agent in plain language: 'launch these 200 creatives, enhancements off, name them {date}_{brand}_{angle}.'
The agent returns the exact ad tree and guardrail results (budget-floor, pixel, link, and placement checks) for approval before publish.
Approve and the engine publishes to Meta. Check status or script the whole flow into your own pipelines via the CLI.
An agent-native ad launcher exposes its core engine to AI agents directly, not just through a web UI. Volume Creatives ships an MCP (Model Context Protocol) server and a CLI, so a media buyer can drive real Meta ad launches from inside Claude, Cursor, or any MCP-compatible tool: planning a dry-run, launching, and checking status with natural-language commands.
The launch engine is the product. The web dashboard and the agent are both thin clients over one launch API. That means anything you can do in the UI (bulk launch, enhancement opt-out, guardrails, dynamic naming) you can do from the agent or CLI, with identical behavior and the same safety checks. Nothing is bolted on as an afterthought.
Whether a launch comes from the UI, the MCP server, or a scripted CLI pipeline, the same pre-launch guardrails run: budget-floor checks for statistical power, missing-pixel detection, broken-link checks, and asset-to-placement matching. The agent cannot skip the dry-run. It returns the plan for approval first, so automation never means losing control.
MCP server
Connect account, plan a dry-run, launch, and check status from any MCP client.
CLI
Script bulk launches and wire them into your own pipelines.
Same guardrails
Budget-floor, pixel, link, and placement checks run no matter how you launch.
Built engine-first
The UI and the agent are thin clients over one launch API.
It is a Model Context Protocol server that exposes the launch engine to AI clients like Claude and Cursor. It provides tools to connect an account, plan a dry-run, launch, and check status using natural language.
Any MCP-compatible client, including Claude and Cursor, plus a CLI for scripting bulk launches into your own pipelines.
Yes. Budget-floor, pixel, link, and placement guardrails and the dry-run preview run identically whether you launch from the UI, the MCP server, or the CLI.
No. The agent returns the dry-run plan (the exact ad tree and guardrail results) for you to approve before anything is published to Meta.
Yes. The CLI lets you wire bulk launches into automated workflows while keeping the same guardrails and idempotent, no-duplicate launching.
Volume Creatives bulk-launches hundreds of Facebook and Instagram ads in one click, with Advantage+ enhancements off, naming applied, and guardrails on by default.