We have deployed version 1.5.0 to the adGPT infrastructure. This marks the deployment of adGPT Studio—our node-based visual execution environment. You are no longer restricted to chat-based commands; you can now construct, deploy, and monitor permanent, background-running automations.
All active workspaces have been updated automatically.
1. The Visual OS (FlowOS Architecture)
We completely overhauled the automation canvas to operate at the speed of thought, allowing you to build complex data pipelines without writing code.
- Tactile Node Canvas: Drag, drop, and connect execution nodes across an infinite, hardware-accelerated blueprint grid.
- Ghost Wires & Synapse Mapping: Connecting nodes automatically maps matching data variables. If you need to manipulate data, visual "Data Pills" allow you to point-and-click to map outputs from one platform (e.g., Google Ads ROI) directly into the inputs of another.
- The Flight Recorder: Every automation runs with complete transparency. A dockable terminal inspector allows you to view raw execution payloads, pinpoint exact failure states, and copy full execution traces for internal auditing.
2. The Architect Copilot (Text-to-Workflow)
Manual assembly of workflows is optional. We have integrated a persistent AI Copilot directly into the Studio environment.
- Generative Blueprints: Describe your required logic in plain English (e.g., "If a Facebook campaign drops below a 1.5 ROAS, pause it and send an interactive approval button to my Slack channel"). The Copilot will mathematically calculate the required logic and instantly construct the wired nodes on your canvas.
- Auto-Healing Engine: If an external API changes or a node fails during execution, the OS will automatically intercept the raw stack trace, translate it into plain English, and suggest a structural fix.
3. Omni-Channel Expansion
The execution engine has expanded beyond strictly media buying. It now connects directly to your agency's operational layer.
- Communications & Approvals: Native integration nodes for Slack and Telegram have been deployed. You can build "Human-in-the-Loop" automations that suspend OS execution until a team member clicks an "Approve" button in a Slack channel.
- Google Workspace & LinkedIn: Connect your ad data directly to Google Sheets (bulk row mapping), Drive, and Docs. Additionally, full LinkedIn integration allows the OS to autonomously manage, publish, and react to professional content.
- Lyria 3 Media Engine: The audio synthesis engine has been upgraded to utilize Lyria 3, allowing for granular BPM and intensity controls when generating multi-modal creative assets.
4. Enterprise Integrity & Background Compute
To support "always-on" automations, the backend architecture was fundamentally restructured.
- Asynchronous Cloud Workers: Your published workflows no longer rely on your browser. They are offloaded to dedicated background worker queues. They will execute silently 24/7, based on the cron schedules or webhook triggers you define.
- AES-256 Vault Encryption: Absolute data sovereignty is critical. All linked platform credentials (OAuth tokens, API keys) are now secured within a stateless AES-256 encrypted vault.
- Circuit Breakers: If an external API goes down, the OS utilizes a 5-strike circuit breaker to automatically pause crashing automations, preventing infinite retry loops and wasted compute.