We have deployed version 1.4.0 to the adGPT infrastructure. This update shatters the Google-only execution boundary, introducing full-stack Meta Ads (Facebook/Instagram) capabilities alongside a revolutionary "Omni-Adapter" architecture that permanently eliminates platform feature lag.
All active workspaces have been updated automatically.
1. The Omni-Adapter (Infinite Execution Scale)
Historically, marketing platforms require developers to manually write thousands of lines of code to support new network features. We have eradicated this bottleneck.
- Universal API Reflection: Instead of relying on hardcoded toolsets, the OS now mathematically reads the entire API architecture of Google Ads, Google Analytics 4, and Search Console in real-time.
- Day-One Feature Parity: If an ad network releases a highly specific, obscure setting tomorrow (e.g., granular Frequency Caps, Content Exclusions, or obscure IP filters), the OS can interact with it immediately. Zero developer waiting time is required.
- Self-Healing Resilience: The Omni-Adapter includes automatic hallucination correction. If the OS drafts a structurally invalid configuration, the reflection engine intercepts the error, auto-corrects the syntax, and deploys safely.
2. Meta Ads Execution Engine
The OS is no longer restricted to Google environments. You can now build, manage, and optimize Facebook and Instagram campaigns directly from the canvas.
- Full Structural Deployment: The engine understands Meta's specific three-tier hierarchy (Campaign > Ad Set > Ad) and can deploy complex configurations, including Advantage+ mandates and detailed demographic targeting.
- Autonomous Creative Hashing: You no longer need to manually host and assign images. When you upload or generate a creative asset, the OS seamlessly exchanges the file for Meta’s required internal Image Hashes and attaches them to your ads automatically.
3. Bayesian Cross-Channel Strategy (Meridian)
Calculating True ROI requires an unbiased view of how platforms interact. We have upgraded the Media Mix Modeling (MMM) engine to support multi-channel data ingestion.
- Federated Ledger Integration: The OS now autonomously aggregates raw spend, impression data, and saturation metrics (Reach/Frequency) from both Google and Meta into a unified data warehouse.
- Causal Inference Math: By breaking down data silos, the Meridian engine can run complex Bayesian math to calculate how your Facebook Top-of-Funnel spend is directly causing your Google Bottom-of-Funnel conversions, eliminating double-counting.
4. Just-In-Time (JIT) Compute Loading
To maintain lightning-fast response times despite the addition of thousands of new capabilities, we engineered a Just-In-Time compute structure.
- Dynamic Tool Summons: The heavy "Universal Discovery" capabilities (scanning massive Meta or GA4 structures) are kept in deep storage. The OS only summons these heavy toolsets when it detects specific strategic intent in your prompt.
- Stateful Cache Invalidation: The OS instantly clears its memory cache the exact millisecond a creation, update, or deletion occurs, completely eliminating "Split-Brain" errors where the OS forgets an action it just took.