← Back to Blog
December 2, 2025•ProductAI Agents

Introducing Campaign IDE: Build Ad Campaigns with AI

We built the advertising equivalent of Cursor. Chat with an AI agent to research websites, plan multi-platform campaigns, validate settings, and launch—all from a single interface.

The Problem: Ads Are Still Built in Spreadsheets

Think about how developers build software today. They have Cursor, VS Code, and AI copilots that understand context, suggest code, and automate repetitive tasks. But marketers? They're still bouncing between:

  • Google Ads interface (one tab)
  • LinkedIn Campaign Manager (another tab)
  • Meta Business Manager (yet another tab)
  • Spreadsheets for budget planning
  • Docs for creative briefs
  • Slack threads for approvals

Each platform has its own UI, targeting taxonomy, and creative specs. Launching a cross-platform campaign takes days of manual work—copying settings, reformatting creative, and double-checking targeting.

The Solution: Campaign IDE

Campaign IDE is a chat-first interface for building ad campaigns. Describe what you want, and an AI agent handles the rest:

› "Create a LinkedIn campaign for Sourcegraph targeting engineering leaders"

The agent researches the website, identifies key value props, builds targeting criteria, generates ad copy, and validates the campaign—all while you watch.

It's the advertising equivalent of pair programming with an AI that knows Google Ads API, LinkedIn Marketing API, and Meta Marketing API inside and out.

How It Works

1. Research Phase

Tell the agent about your product or paste a website URL. It scrapes key pages, identifies your ICP, value propositions, and competitive positioning. This context informs every decision downstream.

2. Strategy Planning

The agent recommends campaign types based on your goals: Sponsored Content for awareness, Lead Gen Forms for direct capture, Sponsored Messages for outbound. It builds targeting criteria, budget allocations, and timeline.

3. Validation (Dry Run)

Before spending a dollar, the agent validates your campaign using dry_run=true. This calls the real platform APIs to check targeting, budget, and creative specs without creating anything.

4. Launch

When you're ready, say "launch it." The agent creates the campaign (paused by default), uploads creative, and sets targeting. One command, all platforms.

5. Performance Monitoring

Ask "how are my campaigns doing?" and the agent pulls live performance data. It surfaces insights, flags underperforming ads, and recommends optimizations.

AI Agent Tools

The Campaign IDE agent has access to specialized tools for each platform:

Performance Tools

  • pull_google_ads_performance — Impressions, clicks, spend, conversions
  • pull_linkedin_ads_performance — Account analytics across all campaigns
  • pull_all_platforms_performance — Unified view across connected platforms

Campaign Creation Tools

  • create_google_ads_campaign — Search, Display, Performance Max
  • create_linkedin_ads_campaign — Sponsored Content, Messages, Lead Gen
  • create_meta_ads_campaign — Awareness, Traffic, Leads, Conversions

Document Tools

  • save_campaign_summary — Save strategy document for sharing
  • update_document_section — Edit specific sections of the brief

All tools support dry_run mode for safe validation before real execution.

Connected Platforms

Campaign IDE works with your existing ad accounts via OAuth. Currently supported:

🔍

Google Ads

Search, Display, PMax

💼

LinkedIn Ads

Sponsored Content, InMail

📘

Meta Ads

Facebook, Instagram

🔎

Microsoft Ads

Bing Search, Audience

🤖

Reddit Ads

Community Targeting

𝕏

X Ads

Promoted Tweets

Connect accounts in Settings → Credentials. All tokens are encrypted and managed via Nango.

Real Example: LinkedIn Campaign in 4 Minutes

Here's a real workflow we recorded. The AI agent:

  1. Researched sourcegraph.com to understand the product and ICP
  2. Built a 3-campaign LinkedIn strategy (Sponsored Content, Lead Gen, InMail)
  3. Validated a Sponsored Content campaign with dry_run=true
  4. Pulled existing LinkedIn Ads performance data
  5. Saved the complete strategy document

Total time: 4 minutes 20 seconds. No spreadsheets. No tab switching. No copy-pasting between platforms.

Watch the demo video →

Coming soon on YouTube: Full walkthrough of building a LinkedIn campaign strategy with Campaign IDE.

Technical Architecture

For the curious engineers:

  • Frontend: Next.js 15, React, TipTap editor with Yjs for real-time collaboration
  • AI: Claude Sonnet 4 via Anthropic API with tool use (function calling)
  • Streaming: NDJSON streaming for real-time tool execution visibility
  • OAuth: Nango self-hosted for secure token management and refresh
  • Platform APIs: Node.js integrations (Google Ads, LinkedIn Marketing v202501, etc.)
  • Database: PostgreSQL via Prisma, encrypted token storage

Read more about our architecture in Building Synter with Amp.

What's Next

Campaign IDE is live today for early access users. We're working on:

  • Creative generation: AI-generated ad images and video concepts
  • A/B test automation: Automatic creative and copy testing
  • Budget optimization: AI-driven spend allocation across platforms
  • Reporting: Natural language performance reports

Try Campaign IDE Today

Join the waitlist to get early access to the future of campaign building. Chat with AI, launch across platforms, and stop switching tabs.

Join Waitlist →Try Campaign IDE