brain bytes lab.PERSONAL PROJECT DEMO

Brain Bytes content pipeline

How a personal AI publishing workflow runs

AI helps research, draft and prepare assets. Evan reviews the evidence and decides what goes live.

Topic

A question worth exploring

Evidence

Sources, claims and known gaps

Draft

Article text and visual assets

Human review

Verify claims and edit the story

Public article

Approve release and check the result

EMBER OUTLINE = HUMAN DECISION POINT

Working today

  • Research briefs and source records
  • Article drafts, diagrams and editorial review
  • GitHub change review and deliberate release

Video work in progress

  • A 59-second narrated explainer, owner-approved
  • Planning profiles for three video formats
  • Longer video formats still pending

Public and synthetic examples only. This personal project implies no employer endorsement.

brain bytes lab.PERSONAL PROJECT DEMO

Article production and review

Articles remain drafts through ready; publication is a separate approved step

A topic becomes a source pack, an outline and an article. The queue records progress; it cannot authorize publication.

Research and authoring

Human editorial gate

research

Topic, sources and open questions

outline

Research brief and a clear thesis

draft

Article text and diagrams

user-review

Fact-check, edit and request revisions

ready

Editorially approved; still a draft

published

Approved promotion and live verification

↺ Revisions return to draft

Evidence review

AI gathers support and flags gaps. Evan verifies load-bearing claims and the final wording.

GitHub review

The change, audit output and build evidence stay reviewable. Checks cannot prove factual truth.

Release approval

Merging main triggers deployment. Owner approval covers that effect, and the live article is checked.

brain bytes lab.PERSONAL PROJECT DEMO

Inside the AI pipeline

The agents, skills and handoffs behind each article and video

HeroLoop · how we build the pipeline

The engineering flow for the system itself: spec, tickets, implementation, review and deliberate release.

IN DEVELOPMENTAutomated runner | Live dashboard

Editorial and media skills · how we produce content

The agent roles, skills and gates below. They turn a topic into a reviewed article and, later, video assets.

researcher

Gathers sources, records claims and flags open gaps

writer

Turns the research brief into an article draft and script

reviewer

Checks claims against sources and flags unsupported wording

media producer

Storyboards, renders and versions visual and video assets

skillEach role runs a skill: reusable instructions that define its inputs, outputs and checks.

Handoffs are saved files
Saved researchDraftsStoryboardsVersioned assets
Execution
Interactive agent sessionResearch, writing, review and storyboarding, with Evan present.
Deterministic rendering and validationRenders, site builds and checks. Same inputs, same output.
Human gates
Source

Evidence is sound

Script

Claims and story

Visual proof

Preview render checked

Narration

Voice and captions

Final output

Watch, listen, verify

Publication

Owner approves release

EMBER OUTLINE = HUMAN DECISION POINT

brain bytes lab.PERSONAL PROJECT DEMO

Two published articles

The same process, each result inspectable with its evidence

Published example

21 September 2026

Jev vs. Laya: Where Decision Models Fit in the SDLC

  • A bounded evaluation using 18 synthetic cases
  • Methods, source links, diagrams and limitations
  • No claim of general reliability or model superiority

Published follow-up

6 October 2026

Where Decision Models Fit in an AI Agent, and Where They Don’t

  • Research brief, claim matrix and source ledger
  • Article, diagrams and a 59-second explainer
  • Any new benchmark remains a plan

One claim should stay connected to its evidence

Source excerpt

What the source actually says

Claim and qualifier

What the article can support

Article and figure

How the reader sees the idea

Open the published Jev vs. Laya article brainbyteslab.org/articles/jev-vs-laya-decision-models
brain bytes lab.PERSONAL PROJECT DEMO

Video extension in progress

The format planner exists, and the first narrated explainer is published

Reviewed article and source pack

Reuse the evidence. Adapt the script, pacing and visuals for each format.

Planning profiles

Shorts

1-60 seconds

9:16 vertical · One clear idea

Briefs

3-5 minutes

16:9 horizontal · Context and recent developments

Focus

5-10 minutes

16:9 horizontal · A deeper topic or walkthrough

First narrated explainer

59 seconds

Owner-approved. Voice, captions and an independent review are done.

Released

6 Oct

Published with the article. Longer formats are not built yet.

Review path · all four gates passed for this explainer

Script review

claims and story

Audio and captions

voice and timing

Full media review

watch, listen, verify

Publish decision

owner and destination

brain bytes lab.PERSONAL PROJECT DEMO

Explainer · 16:9 · 59 seconds

The first narrated explainer

One clear idea from the decision-models draft

  • Script checked against the article's sources
  • Synthetic narration, captions and motion graphics
  • Published with the article, 6 October 2026
Video missingExpected at/assets/articles/decision-models-agent-runtime/explainer.mp4