Have it built
Book 15 minutes with me
Fifteen minutes is enough to see which one fits. I’ll tell you what I’d build first, and whether you need me at all.
Former agency CTO · Led a team of 12 developers · 45+ digital projects managed
AI marketing and content engine · AI presenter
I build the engine behind your content. You approve what goes out.
You know your subject. What you don’t have is a system: a voice guide the AI writes in, four pillars that decide what you talk about, a five-beat script that holds attention, and a keyword path from a comment to a booked call.
I build that system with AI as my team, then hand you the controls. If you want one, an AI presenter of you delivers the videos, with your written consent and a clear AI label on every post.
Theo · the character my content pipeline was built around · Blender render
Why content stalls
Your AI drafts sound generic because it has never read you.
Your drafts sound like everyone else’s.
The AI has never been given your voice, your opinions or your examples, so it writes the average of the internet.
Fixed firstYour voice
You post in bursts.
Nothing decides what you talk about, so client work wins, the account goes quiet, and every restart begins at zero.
ThenYour pillars
People comment, like and save, and nothing happens next.
There’s no keyword, no page, no path to a conversation.
Each of these has a structural fix, and the engine below takes them in this order: your voice, then your pillars, then the path to a conversation. More help without a system just makes more posts.
The engine
One pipeline, from what you know to a conversation booked.
This is the shape of every engine I build. The tools change per client. The stages don’t. Tap a stage to see what it produces.
Produces · Voice guide
Voice guide
Built from your own posts, talks and call transcripts. Each rule comes with an example of you doing it, plus a never-say list, a spoken version for video and a claims policy in three columns: what you may state, what needs proof, what you never claim.
Produces · Pillar matrix
Four pillars
Four pillars, each with a job: Method (how you work), Proof (what you’ve done and can show), Point of view (where you part ways with your field) and Path (the step toward your offer). The default mix is 35/30/20/15, adjusted to the proof you actually have. Every pillar answers a question a buyer has really asked you.
Produces · Proof-gap list
Proof inventory
Everything you can show on screen, with its permission status. What’s missing becomes a proof-gap list in priority order, so no script leans on proof you can’t use.
Produces · Hook bank
Hook patterns
A bank of named patterns and a starter set of five or six that fit your pillars. A pattern is judged only after three uses. Structure only, never anyone’s wording.
Produces · Five-beat script
Five-beat scripts
Hook, tension, proof, payoff, keyword CTA. Claude skills draft them in your voice, and every draft passes a claim check against your claims policy before you see it. You edit and approve.
Produces · Finished video
Production
You on camera, an AI presenter of you, or a rendered character. Captioned, levelled, labelled and checked against the quality gate.
Produces · Ten-asset set
Repurposing
One approved idea becomes ten assets, from the reel to a carousel, a written post and a saved DM reply, each reshaped for its format.
Produces · DM path
Keyword funnel
A comment or story-reply keyword triggers a DM that sends people to one page with a free resource and a next step. I test it from a second account before it goes live.
Produces · Decision log
Weekly review
The sheet compares like with like, reels against reels, and returns one decision a week. It’s logged with a date, so next week changes one thing, not ten.
The free Content Engine gives you v1 of the voice guide, pillars, hooks and scripts. A paid build adds production, the presenter and the funnel wired into your own accounts, then runs the reviews with you.
The stack, by job: Claude runs the drafting skills, ElevenLabs does voice, ManyChat runs the keyword DMs, a Google Sheet holds the calendar and the review. For rendered characters I add Blender, Remotion and ffmpeg, as in the Theo build below. Quality gate on every video: a hook in the first second, a visual change every two to three seconds, a reason to save or send, claims that pass the claim check, a clear sign-off.
What the engine produces
Real files you keep, not a strategy deck.
Here’s the kind of artefact the engine produces. It’s a fictional example, for a fictional client: a five-beat script that has already passed its claim check and its gate check.
Fictional examplescript-run-01.txt
FICTIONAL EXAMPLE · not a real client
Client:a physiotherapist who treats runners
Pillar:Point of view · "Pain is information, not a stop sign"
Buyer question:"Should I keep running through shin pain?"
Offer it feeds:4-week return-to-running programme
Hook pattern:myth vs mechanism · 35–40 s
[0–3 s HOOK]"Calf stretches aren't the whole story for this kind of shin pain. Here's what I check first."
[3–10 s TENSION]You stretch every day and it still flares at the same point in every run.
[10–25 s PROOF]The step-down test I run in clinic, filmed from the side.
[25–35 s PAYOFF]On screen: "Check the load before you add the stretch."
[35–40 s CTA]"Comment RUN and I'll send you the 4-week plan."
Claim check:no cure or outcome claims ✓ · "see a clinician" line in caption ✓
Gate check:hook on screen < 1 s ✓ · visual change every 2–3 s ✓
Proof from the lab · Theo
The engine, built first for a 3D character.
Theo is the trainer character for Misclo, my own fitness and food app, in development and launching 1 January 2027. Theo’s base mesh came from Tripo, an AI 3D generator. I rigged and animated him in Blender with tools my AI wrote, and measured every change against reference art instead of eyeballing it.
Then I built the content engine around him. Theo proves the research, script, voice, compose and gate stages. The presenter of a real person is the next build, starting with mine.
How one reel is built
- Research
A script analyses short videos I approve for study, one batch at a time: pacing, words per second, time to first speech, cut length, loudness, hook type and call to action. They’re used for structure only and are never republished.
- Templates
Four formats drawn from that structure, written from scratch. Rule: copy the structure, never the words.
- Render
Presets for camera, background and glow, rendered as passes so one render serves many edits.
- Voice
An ElevenLabs library text-to-speech voice I chose and locked for Theo. It isn’t my voice and it isn’t a clone. Word timings come back with the audio, so captions sync without guesswork.
- Compose
Captions, labels, arrows and wipes in Remotion, with ffmpeg for the final mix.
- Gates
A storyboard first, then a cut, then my review. My first demo used a placeholder voice and I rejected it. The finished reel, SB-01 v2, is 17.6 seconds at 1080×1920.
Not posted yet, so no reach, views or followers are claimed. The view counts in my research database belong to other accounts and never appear on this site.
Who builds it
A former agency CTO who now builds with AI.
I’m a former CTO of a US marketing agency. I led a team of 12 developers and I’ve managed 45+ digital projects. Today I work solo from Bali, with AI as my team, and I build in public. Misclo, Theo and this website are how I test the method before I sell it.
- 45+digital projects managed
- 12developers in the team I led
- 17.6sone finished reel from the content engine
How the work splits
You bring the expertise and the final word. I build the machine.
You
- Your knowledge, your opinions, your examples
- Approve every script and every final cut
- Own your accounts and decide what gets posted
- Give written consent if you want a presenter of you
Me
- Strategy, the voice guide, pillars and formats
- Build and test the pipeline, set the quality gates
- Wire the keyword funnel and the review sheet
- Run the presenter production and its disclosure
The AI
- Drafts scripts, captions and repurposed versions in your voice
- Voices, captions, levels and assembles, inside the gates
- Transcribes and analyses what you approve for study
- Generates the presenter video from your approved script
How we work
Map, build, launch, then let it learn.
Map
A free 15-minute call, then a written scope: your audience, your offers, what you’ve already made, and which package fits. If the honest answer is “do it yourself with the free tool”, I’ll say so.
Build
I extract your voice, set the pillars, build the hook bank and the Claude skills, and wire the funnel. You see each piece before it replaces the last, one change at a time.
Launch
The first 30 days go into the calendar as drafts you approve. If you chose the presenter, its first batch arrives with labels in place. You get a recorded walkthrough of how to run it.
Improve
Every week the review sheet returns one decision: what to change, and how we’ll know in seven days. Each month we keep what works and replace what doesn’t, on Engine Run or on your own.
Investment
Starting prices, then a fixed scope after a call.
Content
Engine Setup
From $1,200
The full system, built and handed over.
- A voice guide from your own material: rules with examples, a never-say list, a spoken version and a claims policy
- Four content pillars by job (Method, Proof, Point of view, Path), each tied to a real buyer question and an offer
- A proof inventory and a proof-gap list in priority order
- A hook-pattern bank with your starter set
- Claude skills for five-beat scripts, captions and repurposing, set up in your own account
- A 30-day calendar with first drafts in your voice, sized to how many pieces you can actually review
- A keyword DM funnel (comment and story reply) to one page and one free resource, tested end to end
- A review sheet that returns one dated decision a week
- First builds
Content
Engine + AI Presenter
From $2,200
The system plus a consented, labelled AI presenter of you.
- Everything in Engine Setup
- An AI presenter of you, made only with your signed likeness and voice consent
- A production pipeline from approved script to finished, captioned, labelled video
- Your disclosure policy: the label wording, where it appears and who approves it
Content
Engine Run
From $600/month
3-month minimum
Monthly production and review on the engine.
- Production to a volume we agree in writing
- The weekly decision, and a monthly review of what to keep, cut and test next
- Pillar, format and voice updates as the account learns
- You own the voice guide, skills, scripts, videos and accounts.
- Starting prices cover a typical scope. Your exact price is fixed in writing before any work begins.
AI presenter · consent and disclosure
An AI presenter of you is only as good as the trust behind it.
An AI version of a real person is powerful and easy to misuse. So the rules come first, they go into the contract, and they apply to my own AI presenter too. Read how I disclose AI
- Only youI make a presenter of the person who signs the consent,
never of anyone else, and never of a public figure.
- Written consentLikeness and voice, for named uses, on named channels.
You can withdraw it, and production stops.
- Always labelledOn the profile, in every caption,
and with the platform’s own AI label wherever it offers one.
- No fake proofThe presenter never presents invented testimonials,
results or claims you couldn’t make yourself.
- Your approvalYou approve every script and every final cut
before anything goes out.
Consent signedDateChannels
Free · keyword CONTENT
Build v1 of the engine yourself, today.
The free Content Engine is the Day 3 tool of AI Business Build Week: a Claude skill plus a pillar, hook and repurposing system. Feed it ten of your own posts and one sentence about your offer. No email wall.
You get
- A voice guide with a claims policy
- A four-pillar matrix
- A hook bank
- Five-beat scripts
- A keyword DM path
Questions
Straight answers.
Will this grow my account or bring me clients?
I can’t promise that, and nobody honest can. What I build is the system that makes consistent, on-brand content possible and gives every post a path to a conversation. What it returns depends on your offer, your niche and how long you run it. No results, reach or income are promised at any price.
Do I have to use an AI presenter?
No. Engine Setup works with you on camera, voice-over only, carousels or text. The presenter is an option for people who’d rather approve than film.
Will people know it’s AI?
Yes, because it’s labelled. Every post with an AI presenter carries the label on the profile, in the caption and through the platform’s own AI tag where there is one. That’s a condition of working with me.
Is the AI presenter my real voice?
It can be, with your written consent for named uses, and you can withdraw that consent. If you’d rather not use your own voice, a library text-to-speech voice works too, checked against the provider’s commercial terms for your plan.
Will it sound like me or like generic AI?
Like you, because the voice guide is built from your own material and every draft is checked against it. You approve each script. Anything that doesn’t sound like you gets cut, and the guide gets updated.
What results do you have?
None for clients yet, and I won’t invent any. What I can show is the method: the Theo pipeline, stage by stage, and the finished reel it produced. When there are results, they’ll be here with a date and a source.
Who owns everything?
You do: the voice guide, the skills, the scripts, the videos and the accounts. If we stop working together, you keep running it.
Do you post for me?
By default, no. You post, or you approve each post before it’s scheduled. Nothing goes out under your name without your OK.
What will the tools cost me?
The stack, by job: Claude for the drafting skills (on a plan with Skills enabled), ElevenLabs for voice, ManyChat for keyword DMs, a Google Sheet for the calendar and review. When I checked on 9 November 2026, Claude’s paid individual plans were around $20/month and ManyChat’s paid plans started around $15/month; check current pricing before you buy. Everything is billed to your own accounts, and your scope lists each tool with its current list price before you sign.
How long does setup take?
The gating input is yours: ten of your posts (or a transcribed voice note) and one sentence that says what you sell. Once I have those, the voice guide and pillars come first. You get a dated plan for the rest in your scope after the call.
Are there other ways in?
If you’re starting a business from zero, the Solo Launch Build ($990) includes a 30-day content plan. It starts on the same free call.
Can I start without paying?
Yes. Take the free Content Engine, or join AI Business Build Week and build it live with me on Day 3. Book a call when you want it built for you.
Next step
Want the engine built for you, or want to build it yourself?
Start on your own
Build it yourself.
Take the free Content Engine and get a voice guide with a claims policy, a four-pillar matrix, a hook bank, five-beat scripts and a keyword DM path. No email wall.
Prefer email? contact@labgenz.com