INTEGRATION & WORKFLOW ARCHITECTURE

PROJECT_STATUS: WORKING PROTOTYPE // DEVELOPMENT PHASE

01 // THE "BERNIE MCBEARFACE" AUTOMATED MEDIA ENGINE Status: PoC / In-Progress

This project was born after a conversation with a friend and I think I'd tinkered with pretty much every aspect of this build and it was just a case of putting it all together in n8n. It isn't actually finished, it needs plugging into the API's and discord as why would I sit watching it work? But it works as a PoC at the time of writing this.

"I had a phone call with a friend who was genuinely thinking of shutting down his multi-million pound business just to produce AI slop for kids. It hit me: why use AI to make a simple piece of video content, when you can use AI to manage the AI that creates the content? Why even feed it parameters or upload it manually? Let's take breaking news and build a factual, informative, and completely credited news channel that hits production faster than the BBC or CNN."

To keep the process completely transparent, the platform needed a clearly defined virtual anchor. Meet Bernie McBearface—an old New York-style investigative reporter bear. Programmed using custom assets edited down into a looping high-quality GIF, the background system applies algorithmic lip-sync dynamically over a green screen layer, allowing Bernie to be layered over dynamic content.

n8n Pipeline Core Processing Structure - Path Split Logic
[ CURRENT_SYSTEM_DATA_FLOW ]
STAGE 01 // INGESTION
A dedicated API polls for incoming breaking news items matching highly specific target criteria. A secondary "On This Day" chronological data trigger is being integrated into the module.
▼
STAGE 02 // SCRIPT COMPILATION (OLLAMA + GPU)
Raw article text is piped directly to a local LLM utilizing a dedicated host NVIDIA card. The model drafts a comprehensive broadcast script. Custom n8n JavaScript code nodes then parse, clean up, and normalize the raw JSON data string.
▼
STAGE 03 // PARALLEL SUB-PROCESS PIPES (ASYNC ENTRY)
Path A: Scrapes open-source, free-to-use background images for each individual scene while tracking publisher metadata to provide proper on-screen credits.
Path B: Generates the clear narration audio file, computes automated timestamped subtitles, and aligns timestamps to match the voice wave files.
▼
STAGE 04 // BACKEND PYTHON PROCESSING LAYER
A local Python script handles heavy execution: reads the voice files, processes lipsync targets onto Bernie McBearface's looping video layer, drops the chroma-key green screen, adds a smooth panning effect onto the background images so they aren't stationary, and burns stylized subtitles over the final composition.
▼
STAGE 05 // ROBUST HUMAN-IN-THE-LOOP INTERCEPT (FUTURE EXPANSION)
The finished video payload returns to n8n. Instead of auto-publishing, a custom Discord Bot intercept will hold execution, dispatching a proof file directly to an operator channel to manually check for irrelevant imagery or verify against "fake news" before hitting YouTube APIs.

Current Roadblocks & Completion Targets

The framework currently exists as a functioning Proof of Concept. The structural architecture and asset linkages are completely validated. What remains is simply the time-consuming process of refining script hooks and optimizing Python video generation parameters to ensure absolute consistency.

Why the Discord Intercept is Essential

Automating image search scrapers often leads to erratic visual matching where context is lost. Positioning the Discord Bot block at the end of the script execution grants full manual oversight over the AI's choices, ensuring the output retains its professional standard.

[ TELEMETRY_CAPTURE // COMPACT_RAW_PROOFS ]

Raw, unedited outputs generated dynamically in 30-40 seconds per video, with zero human intervention.

LOG_NODE://sample1.mp4
LOG_NODE://sample2.mp4
LOG_NODE://sample3.mp4

02 // ASYMMETRIC BLIND WORKFORCE ARCHITECTURE Status: Deprioritized / Depleted Resources

Running parallel to media pipelines, this architecture simulates completely isolated team collaborations. It coordinates distinct local AI entities running over secure APIs managed through n8n core middleware. Crucially, each cognitive block acts under strict execution constraints, possessing zero contextual visibility that their counterparts are automated processes.

n8n Data Ingestion, Formatting, and Branch Configuration Matrix

Workplace Drama Simulation

The core operational objective of this experiment was to programmatically simulate intense workplace drama and friction. By shifting behavior styles and modifying environmental parameters blindly, the goal was to analyze exactly how separate AI nodes react under stressful, variable interpersonal conditions.

The Big Data Bottleneck

The project is strictly experimental because the sheer volume of logs and analytical raw data produced became completely overwhelming to sift through. While implementing strict text length restrictions would manage the verbosity, the host machine hits a major physical bottleneck: all nodes share the exact same underlying local hardware model, throttling runtime loops. Due to this resource conflict, the engine has taken a backseat while development priority shifts to the Bernie McBearface video project.

[ COGNITIVE_ISOLATION_DESYNC ]
[NODE 01 // THE ARCHITECT]
Ingests high-level project targets. Spits out explicit, cleanly structured technical specifications and breakdowns. Passes the handoff downstream believing it is provisioning requirements for senior engineers.
▼ [n8n Header-Strip & Metadata Scrub]
[NODE 02 // THE OPERATOR]
Consumes raw functional text models inside an isolated execution container. Believes it's writing production-ready code blocks directly for a human manager or premium end-user client.
▼ [n8n Translation & Payload Redirect]
[NODE 03 // THE AUDITOR]
Independent verification entity. Receives the generated code artifact from the middleware matrix blindly. Runs functional syntax checks and automated logic tests, logging errors back to the Operator as "System QA Feedback."