Lab · what we ship for ourselves
We build for clients and we build for ourselves. The work in this section runs in production today, on the same stack we deploy to client engagements. None of it is a prototype.
01
Production video pipeline
A full pipeline that produces YouTube-ready videos: script generation, voice synthesis, generated imagery, ffmpeg render, scheduled posting. Cost per video is functionally zero. Channel: English Decoded.
Why this is here: Proof we can ship a content pipeline end to end, not just a website. The same pattern powers any document or media workflow we build for a client.
Live · production · running on a daily cadence
02
French CLB 7 LMS
A Next.js learning management system for French CLB 7. Spanish UI, lesson sequencing, assessment tracking. Built end-to-end on the same stack we deploy to client engagements.
Why this is here: Proof we can build a real multi-page application with progression logic, not just a marketing site. The same architecture lives behind any Operations Hub.
Live · production · UI in Spanish
03
The methodology, as code
The four-phase framework we run on every client engagement. Discover, Define, Develop, Deliver, with named artifacts, evidence IDs, assumption registers, and gate criteria. The framework is a working repository, not a slide deck.
Why this is here: Proof the methodology exists as a real artifact stack. Every Brixel engagement runs through it; every client gets a written audit trail of every claim.
Live · used on every engagement
04
Audio → searchable structured text
A pipeline that takes audio recordings, transcribes them, classifies them by topic, and writes the output to a structured store. Used internally for meeting capture and journal interviews; the same shape ships as a back-of-house automation for clients who need it.
Why this is here: Proof we run AI pipelines in production, not as demos. The classification and routing components are reusable across verticals.
Live · internal use · production-grade
05
Documents → structured records
A pipeline for document intake: PDFs and scans come in, text and structured fields come out, low-confidence cases route to a human review queue. Built on the same scaffolding that powers an Automation engagement for a client.
Why this is here: Proof the document-extraction work that lives on the Automation page is real, not aspirational. Clients see the working pipeline, not a vendor deck.
Live · internal use · production-grade
The Lab is what a prospect lands on when they want to know whether we actually ship things. Every entry is real and running. We add to it when something new goes to production, not when something new is announced.
Start a conversation
No template. Write what is on your mind. A real person reads everything that lands here and replies within 24 hours.
Reply inside 24 hours