Content Playbook

What to publish, where, at what cadence, in what aesthetic. Four platforms, one brand: Building an AI Platform in Public. Repurpose one experience across all four — don’t create four different pieces.

The mix is 80% tech / 15% lifestyle / 5% faith (see Identity). The production aesthetic is voice-over + cinematic + system-as-protagonist. No talking head.

This playbook is the operational layer on top of Identity. Identity says what the brand is. This doc says how to publish it — which platform, what cadence, what format, what aesthetic, what NOT to publish. Five years of consistent content production run on patterns documented here, not on willpower.

The single most important rule: one experience produces five pieces of content, not one. The ROOT phase work is the only renewable resource. Everything else — the YouTube long-form, the LinkedIn lessons, the Instagram carousel, the Facebook update, the optional blog post — cuts from a single experience. If a week’s content requires five separate experiences to fuel it, the cadence is already broken.

Why repurposing is the load-bearing discipline

Most “build in public” creators burn out within 12 months because they treat each platform as a separate content pipeline. Four pipelines × three months = exhaustion. ROOT runs five years. The math only works if one experience cuts five ways. The “I deployed Kubeflow” worked example below is the canonical pattern — internalize it before publishing anything.

---

The four platforms: priorities

1. LinkedIn      Professional home. Build-in-public updates, lessons learned, architecture
                 diagrams, career transition content. Highest priority for AWS Community
                 Builder + CNCF Ambassador credibility.

2. YouTube       Long-form content library. The journey documented in detail.
                 Tutorials, deploy walkthroughs, year-end retrospectives, launch demos.

3. Instagram     Visual storytelling. Reels for daily progress, carousels for architecture
                 explainers, stories for behind-the-scenes. Reaches tech-shifters audience.

4. Facebook      Community + personal connection. Cross-post, journey updates, photos,
                 personal stories. Lower priority, but Filipino-Christian-tech reach matters.

---

Cadence

Year	YouTube	LinkedIn	Instagram	Facebook	Blog	Notes
Y1	1-2 / month	3-4 / week	3-5 / week	cross-post	2-4	Build cadence quietly. Audience grows from short-form.
Y2	1 / week	3-5 / week	3-5 / week	cross-post	6-10	terralabs launch. Then 1/week cadence locks in.
Y3	1 / week	3-5 / week	3-5 / week	cross-post	10-12	basecamp public launch. Year-end retro.
Y4	1 / week	3-5 / week	3-5 / week	cross-post	8-12	Quiet build year for llm-gateway.
Y5	1 / week + 4 launches	4-6 / week	4-6 / week	cross-post	10-15	THE year. Studio + mlship + grad.
Total	~250 videos	~800-1000 posts	~1000+	cross-posted	~36-50

The non-negotiable: 1 YouTube long-form per week + 3-5 LinkedIn posts per week. That’s the foundation. Instagram is daily-ish but light effort (cuts from the YouTube + photos). Facebook is largely cross-posted from LinkedIn or Instagram.

Cadence gaps compound

Missing one Sunday is a dent. Missing two in a row is a pattern. Missing three is a quit. Both the algorithm and the audience reward continuity — every gap forces re-introduction, and the flywheel slows. The 1-YouTube + 3-LinkedIn floor is the minimum that keeps it turning. Below that, growth flatlines and you’re publishing into a void. Heavy ROOT phases (Y4 P25, Y5 P30 capstone) are the only sanctioned exception — and they get pre-announced cadence cuts, not silent gaps.

---

Repurposing: one experience, four platforms

This is the operational core. Don’t create four different pieces of content. Repurpose one experience across all four.

Worked example: “I deployed Kubeflow on Kubernetes” (Y4 P23)

Platform	Format	Length	Effort
YouTube	”Deploying Kubeflow on Kubernetes — a 15-minute walkthrough”	15 min	8-12 hrs (record + edit + upload)
LinkedIn	”Lessons learned deploying Kubeflow — 3 things that surprised me” + screenshot of pipeline DAG	200-400 words	30 min (extract from video script)
Instagram	Carousel: “Day 15 of Building an AI Platform in Public — Kubeflow.” 5 slides: What is Kubeflow? Why I chose it. The deploy. Challenges. Next step.	60-90s viewing	1 hr (Canva or Figma)
Facebook	Personal-update post + a photo from the desk. “Another milestone in the AI Platform journey…“	1-2 paragraphs	15 min (lightly adapt LinkedIn)
(Optional) Blog	”Deploying Kubeflow — what I’d do differently” — 800-1500 words	1-2 hrs	If the experience has a deeper lesson worth a long-form post

One Kubeflow deploy = 5 pieces of content. ~10-15 hours total instead of 5x that.

Repurposing rules

1. The YouTube long-form is the source. Everything else cuts from it.
2. LinkedIn = lessons learned, no fluff. Specific tech takeaways, no aspiration.
3. Instagram = visual. Carousel for explainer, Reel for short clip, Story for daily progress.
4. Facebook = personal-update tone. Slightly warmer than LinkedIn; same content, different tone.
5. Cross-post timing: YouTube Sunday → LinkedIn Mon-Wed lessons → Instagram carousel mid-week → Facebook update later in week.

---

Production aesthetic

References (steal from):
  Apple product films       hero-shots-of-the-system; voice-over; intentional pacing
  Tulio Sasaya              production aesthetic, B-roll, floating UI overlays, subtitles
  Werner Herzog             documentary tone (heard, not seen)
  Cleo Abram                cinematic tech storytelling without creator-bro energy

Avoid:
  Sponsor-driven tech-bro luxe (the Cursor-paid Tulio version)
  Lifestyle aspiration without substance
  Preachy / proselytizing faith framing
  Talking-head explainer (Fireship-style direct-to-camera)
  Prompt-engineering / "10 AI tools" content

Tools:
  Camera:        iPhone (latest) on tripod / gimbal; occasional second angle
  Mic:           lavalier OR shotgun mic for clean voice-over
  Editing:       Final Cut Pro OR DaVinci Resolve (free) on the MacBook
  Motion:        After Effects OR Final Cut for floating UI overlays
  Subtitles:     auto-generated then hand-corrected; muted color (NOT amber-cute)
  Music:         optional, sparse (launch videos yes; daily content usually no)

Voice:
  Voice-over narration. Calm, specific, demo-driven. Your own voice (not synthetic).
  ESL accent is authenticity. Re-record any take that didn't land.
  No on-camera talking head.

The “Apple product film” reference, decoded

System as protagonist means: when the camera moves, it moves over the terminal, the Kubernetes dashboard, the Iceberg query plan — not over a face. Voice-over means: every word is recorded after the visual is locked, never on the fly. Intentional pacing means: no frantic cuts, no “wait — guess what happened next?!” energy. The brand is durable because the aesthetic refuses to chase short-form trend cycles. Five years from now the cuts will still feel right; cuts that mimic 2026 TikTok energy will feel dated by 2027.

---

Video genres (mapped to the 4 platforms)

1. YouTube long-form: “Day N of Building an AI Platform in Public” or topic walkthrough

Length:    8-15 min (sometimes 20+ for big tutorials)
Cadence:   1 / week
Examples:
  - "Day 15 — Deploying Kubeflow on Kubernetes"
  - "How I built the Iceberg lakehouse on a $300 mini PC"
  - "My MLOps architecture on AWS — full walkthrough"
  - "Building a RAG Pipeline from scratch"
  - "AI Platform Roadmap — the 5-year arc"

Structure:
  Cold open    — outdoor / lifestyle B-roll, 5-10 sec, hooks the viewer
  Setup        — what we're building today + why it matters (1 min)
  Walkthrough  — the actual technical work, multi-cut (5-10 min)
  Lessons      — 3 specific takeaways, called out clearly
  Close        — outdoor / sunset / reflection, CTA

Tone: cinematic-but-specific. Voice-over throughout. No talking head.

2. Instagram Reel / YouTube Short / TikTok: short-form daily-rhythm content

Length:    60-90 seconds
Cadence:   3-5 / week
Examples:
  - "Day 15 of Building an AI Platform in Public — Kubeflow"
  - "3 things Postgres replication taught me this week"
  - "What's running on my homelab today"
  - "Sunday weekly log — Y2 Phase 9"

Structure:
  0:00-0:10   Cold open + hook
  0:10-0:55   The work + 1-2 specific points; floating UI overlays
  0:55-1:30   Close + CTA

Tone: cinematic + lifestyle B-roll wrapping; tech is the substance.

3. Launch video: Apple product film aesthetic

Length:    3-5 min (mlship), up to 10 (graduation)
Cadence:   6 across the program
Examples (per launch-playbook.md):
  Y2 P9    — terralabs walkthrough (3 min)
  Y3 P19   — basecamp tour (5 min)
  Y4 P24   — notes-rag in action (3 min)
  Y5 P29   — Abukix Studio tour (5 min)
  Y5 P30   — mlship v2 demo (3 min)
  Y5 grad  — "5 years in" (5-10 min)

Aesthetic: Apple product film — system as protagonist, voice-over, intentional pacing,
           optional subtle music, NO face on camera.
Production: 8-15 hours per video. See launch-playbook.md.

4. Year-end retrospective: Werner Herzog documentary aesthetic

Length:    5-10 min
Cadence:   1 / year (Y2 onward)
Voice-over over montage from the year — code commits, weekly log entries, homelab evolution,
run / sunrise / Sunday shots. The rhythm rendered visible. Closes the year emotionally.

---

Instagram carousel format

The highest-leverage Instagram format for explainer content. 5-10 slides per carousel.

Slide 1: hook — "Day 15 of Building an AI Platform in Public"
Slide 2: what is X (the thing you deployed/learned)
Slide 3: why it matters
Slide 4: how it works (architecture diagram or screenshot)
Slide 5: what surprised me / lesson learned
Slide 6: what's next
Slide 7: link in bio + abukix.dev

Reuse the architecture diagrams from blog posts. Reuse screenshots from YouTube videos. Reuse lessons from LinkedIn posts. One experience, multiple cuts.

---

Blog post genres

1. The journey post

Title format:    "X at end of Year N — what I learned"
Frequency:       Once per year (Year-end retrospective) + once per major launch
Length:          1500-3000 words
Examples:
  - "terralabs at launch — multi-cloud IaC the side-by-side way"  (Y2 P9)
  - "basecamp at end of Year 3 — what I learned"                  (Y3 P19)
  - "Abukix Studio at launch — 5 years of platform engineering"   (Y5 P29)
  - "ROOT graduation — 5 years in"                                (Y5 P30)

2. The pattern deep-dive

Title format:    "X vs Y: I built both for N months. Here's what's different."
Frequency:       2-4 / year (when a pattern entry hits DEEP)
Length:          2000-4000 words

3. The sanitized postmortem

Title format:    "How my homelab X went down at 3am, and what I rewrote"
Frequency:       0-2 / year
Length:          1500-2500 words

4. The “small useful” post

Title format:    "Why I picked X over Y for $SPECIFIC_CASE"
Frequency:       Filler
Length:          600-1200 words

---

What NOT to publish

The 1% that breaks the brand

Most of this list is “off-brand” — survivable. Three rows are brand-breaking if they ship: AI-generated motivational content (cheapens every other post retroactively), preachy faith content (collapses the 5% guardrail), and prompt-engineering / “10 AI tools” content (puts the brand in the saturated creator lane it deliberately avoids). When in doubt, the publish-rejection cost is always lower than the brand-recovery cost.

Don’t	Why
Tutorial-style “how to install X” with no insights	Saturated; not differentiated
Hot-take opinion pieces	Burns trust
AI-generated motivational content	Worse than not posting
Posts about industry drama	Off-brand
Long marketing-speak (“robust”, “scalable”, “leverage”)	Engineering writing has cold-take energy
”10 AI tools you must use” / prompt-engineering content	The saturated lane; not your brand
Sponsored content that overrides the brand	Sponsors may come; the brand stays
Faith content that’s preachy or proselytizing	Faith is 5% present, never the subject
Posts during heavy phases (Y4 P25 / Y5 P30 capstone)	Capstone work > content production

---

LinkedIn templates

Build-in-public update

Day N of Building an AI Platform in Public.

Today I deployed [thing]. Here are 3 things that surprised me:

1. [specific tech insight]
2. [specific tech insight]
3. [specific tech insight]

Architecture: [link to blog or screenshot]
Code: [GitHub link]

Launch post

Just shipped X.

[1-sentence value prop]

What it does:
- bullet
- bullet
- bullet

Why I built it: [1-2 sentences, NOT marketing]

Read more: [link to blog post]
GitHub: [link]

Milestone post

[N] months into building AI platforms in public.

What's true now that wasn't true 6 months ago:
- specific
- specific
- specific

What I'm still figuring out:
- specific

[link to a journey post for depth]

---

Cross-references

• Identity — the brand the content serves
• Launch Playbook — when to launch what (with multi-platform repurposing)
• Hosting Plan — technical surface YouTube/blog/Studio live on
• Accounts Setup — handles + domains
• The Story — the rhythm content production maps to
• Master Plan — the curriculum the content documents
• Blog Post Template — the post template
• Weekly Log Template — Sunday write-up format
• Studio composition recipes — the killer demos that anchor each loud launch