Writing
How we build and run the site and the media.
On the system, the media, brand voice, and the Claude Code tools we publish.
LatestOne Timeline: How Voice, Captions and Sound Land on the Same FrameSep 29, 2026
Start here
Generation Got Free. Installing It Didn't.
AI made content generation nearly free. It did not touch the hard part: the judgment to know what to make, the context to make it sound like you, and the work of wiring it into a real business and running it. That last mile is the deployment layer, and it is where all the value moved. This is the case for installing a brand system instead of buying more disconnected content, how it compares with an agency or an in-house hire, and how the Sprint, the Setup and the Retainer put it in place.
One Founder Video In. A Month of On-Brand Content Out.
Two founders, two very different brands, one workflow: they record a single video on their phone and get back a full run of captioned reels, posters, and carousels in their own brand style. Here is how the content system does it, and why the speed comes from infrastructure, not shortcuts.
AI Slop: What It Is, Why It Happens, and How to Fix It
AI slop is generic, low-quality, AI-generated content produced at scale with no editorial standards. Here is what it looks like, why it happens, the 7 signs to spot it, the 5 patterns that fill LinkedIn, and the infrastructure that fixes it.
Brand as Code: How Founders Use Claude Code to Stop AI From Drifting Off Their Voice
Most founders re-explain their brand voice to AI every single conversation. Six months ago I wrote my brand into a Claude Code config file once, and every tool I run now reads it before producing anything. No drift, no slop, no re-explaining. This is the long-form companion to Episode 1 of Claude Code for Founders. The mechanics of brand as code, the Claude Code skill stack that composes into a brand operating system, and how the whole thing scales social media output across every channel without losing voice consistency.
Generation Got Free. Installing It Didn't.
AI made content generation nearly free. It did not touch the hard part: the judgment to know what to make, the context to make it sound like you, and the work of wiring it into a real business and running it. That last mile is the deployment layer, and it is where all the value moved. This is the case for installing a brand system instead of buying more disconnected content, how it compares with an agency or an in-house hire, and how the Sprint, the Setup and the Retainer put it in place.
Your Website, Your Socials, and Your Reviews Are One Machine. Most Businesses Run Them as Three.
Your website, your social channels, your Google reviews, and your video are not four separate chores. They are one loop: media gets you seen, reviews get you trusted, the site gets you chosen, and a job done well feeds all three. This is the pillar guide to wiring that loop into one machine: voice captured as rules, one shoot cut into a month of content, publishing on rails, reviews asked for every time, and a smart website that search engines and AI assistants can actually read.
You Are the Ceiling of Your Own Company.
Every founder-led brand hits one wall first: the founder becomes the bottleneck. The team needs you, the audience needs you, the calendar needs you. The problem is not effort. It is architecture. Here is how it shows up, the four ways founder content fails, and what it takes to capture a founder's voice so the brand keeps shipping when the founder is busy.
One Timeline: How Voice, Captions and Sound Land on the Same Frame
Voice, captions, graphics and sound cut to one timeline, so every word lands on its frame, at any length, in every size and hook you want to test.
One Founder Video In. A Month of On-Brand Content Out.
Two founders, two very different brands, one workflow: they record a single video on their phone and get back a full run of captioned reels, posters, and carousels in their own brand style. Here is how the content system does it, and why the speed comes from infrastructure, not shortcuts.
Raw Phone Footage at 8 AM. A Live Campaign Funnel by Lunch.
Session One of a live session series ended Monday at 1:30. The next morning the founder's raw phone footage landed in a shared folder: five clips, a twenty-nine minute session recording, eight photos, one notes doc. Under two hours later the first captioned, brand-typed clips were ready to post. By early afternoon the full kit was live: thirteen finished videos, a carousel set, a funder one-pager, a private delivery hub, a recap article on his site with the film and a data diagram, a Happening Now section on his homepage, and RSVP routing straight to his inbox. This is what the deployment layer looks like when a real community event needs to become a campaign.
They Booked a Barbershop for a Day. The Campaign Shipped That Same Night.
A recovery training company is launching a live session series, a recovery movement that starts in the barber's chair. I shot the whole session in a barbershop on a Canon R5, handed the raw footage to the system, and it organized, cleaned, and aligned everything against the event's funnel page, turning one shoot into a full launch kit (a hero film, message reels, shop reels, posters, and quote cards) plus a four-week posting plan for July 20.
Programmatic Video Kills the Revision Loop
Most branded video goes through weeks of revision rounds over fonts, colors and logo size. Programmatic video renders brand-locked output from a config file. Same input, same result, every time.
Brand Voice
Voice written down as rules, how it drifts, and how to keep AI from producing slop.
Run the free scanYour AI Sounds Generic. Here's the One-File Fix.
AI content sounds generic because you re-explain your brand from scratch every time, and the tone drifts a little more with every prompt. The fix is to make your brand voice a file, not a vibe: forty rules a machine can follow, loaded once and enforced on every post. This is the long-form guide to brand voice as code, why AI tone drift happens, what goes in a brand voice file, and how to make AI sound exactly like your brand at scale.
Views Are Vanity, Audience Is the Asset: Stop Making Content That Disappears
Most content is engineered for a dopamine spike: fast cuts, loud color, non-stop motion. It earns the view and then evaporates. A view by itself is worth nothing; the people who stay are the entire game. The long-form companion to the WhyStrohm short on why views are a vanity metric, what an audience actually is, and how to build a content system that compounds instead of one that disappears. Define your voice once, then let the system run without you. Built, not winged.
Brand Voice Drift: The Measurable Problem AI Made Worse
Your homepage doesn't sound like your LinkedIn. Your LinkedIn doesn't sound like your email. That gap is brand voice drift, and AI tools sped it up. Drift is measurable: vocabulary overlap, cadence, structure, tone and forbidden-word violations. This is the pillar post on how to diagnose and prevent it: the scales we calibrate, the four drift patterns, why a style guide cannot stop them, the five-layer enforcement model, and the three controls that hold a voice in place.
The Founder Fingerprint: Why Your Voice Is Your Brand's Only Moat.
Every founder has a fingerprint. Not your logo. Your voice. Your eye. Your edge. As you scale, it smudges. The handoff smudges it. The AI averages it. The hire guesses at it. By post 100, nobody can tell it's yours. This is the founder tax nobody warned you about, and the system that cuts it. The Fingerprint is the WhyStrohm mission film. Founder-led companies know this pain. Most can't name it. Watch the 55-second mission, then read the full breakdown of voice extraction, brand infrastructure, and why founder identity is the only thing that does not commoditize at scale.
AI Slop: What It Is, Why It Happens, and How to Fix It
AI slop is generic, low-quality, AI-generated content produced at scale with no editorial standards. Here is what it looks like, why it happens, the 7 signs to spot it, the 5 patterns that fill LinkedIn, and the infrastructure that fixes it.
Shotkit and Claude Code
Shotkit, and the Claude Code for Founders series: the file, the memory, skills, MCP and the loop.
ShotkitShotkit 3.1: Plan a Video in Claude Code, From Brief to Checked Frames
Shotkit 3.1 is a free, open-source kit for Claude Code that plans a video: storyboard, prompts for ten generators, and a verdict on every frame.
You Don't Have a Content Problem. You Have a Pre-Production Problem.
Most founder-led brands burn weeks on AI video output that never feels on-brand. The bottleneck is not the generator. It is upstream. Pre-production is the load-bearing layer that turns vague briefs into shots, prompts, and previews you can actually ship. Shotkit is the free, open-source kit for Claude Code that runs that layer. This post walks through the diagnosis (vague brief plus model roulette equals brand drift), the method (a five-layer prompt anatomy with a frozen brand-lock snapshot), and what the five Shotkit skills produce, updated for Shotkit 3.1.
Ritual: A Claude Code Skill That Reads Your Shell History and Drafts Your First Scheduled Trigger.
Claude Code routines shipped. Most people have no idea what their first one should be. Ritual scans your shell history, git repos, and Claude Code memory, ranks your top 5 routine candidates, and drafts a paste-ready trigger prompt from your actual patterns. Open source, MIT.
Claude Code for Founders, in episode order
Brand as Code: How Founders Use Claude Code to Stop AI From Drifting Off Their Voice
Most founders re-explain their brand voice to AI every single conversation. Six months ago I wrote my brand into a Claude Code config file once, and every tool I run now reads it before producing anything. No drift, no slop, no re-explaining. This is the long-form companion to Episode 1 of Claude Code for Founders. The mechanics of brand as code, the Claude Code skill stack that composes into a brand operating system, and how the whole thing scales social media output across every channel without losing voice consistency.
The CLAUDE.md File That Locks Your Brand Voice (Full Anatomy + Copyable Template)
Your brand voice gets paraphrased from memory every quarter, and each time it drifts further into corporate slop. The fix is one markdown file that every AI tool reads before it produces anything: voice, tone, forbidden words, required structures. This is the long-form companion to Episode 2 of Claude Code for Founders. The full anatomy of the file, why brand voice drifts without it, how every tool downstream reads the same spec, and a copyable CLAUDE.md starter template you can drop into your own repo today.
The MEMORY.md File: How Claude Code Remembers You Between Sessions
Every other AI conversation starts with a stranger who forgot you. Claude Code does not. It writes down what mattered at the end of a session and reads it back before the next one. The long-form companion to Episode 3 of Claude Code for Founders. Full anatomy of the MEMORY.md file, why context drift kills AI workflows, how memory gets written and recalled across sessions, why it beats vector stores and chat history, and a copyable MEMORY.md starter template.
What Are Claude Code Skills? How to Turn a Repeated Workflow Into One Command
A skill is a workflow you stop re-explaining. You teach Claude Code the steps once, give it a name, and from then on a single line of input runs the whole thing. The long-form companion to Episode 4 of Claude Code for Founders: what a skill actually is, why re-explaining the same process every time is a hidden tax on your work, how skills compose into a brand operating system, and how to build your first one from a workflow you already repeat.
What Is MCP? How to Connect Claude to Your Live Business Data
MCP (Model Context Protocol) is the open standard that connects Claude directly to the tools your business already runs on, so it reads your live data from the source instead of a stale snapshot you pasted into a chat. The long-form companion to Episode 5 of Claude Code for Founders: what MCP is in plain English, how the host, client, and server fit together, how Claude calls a tool and reads a live source, the control model that keeps you in charge, and how to connect your first tool.
What Is the Loop? How Claude Code Runs the Work Without You
An AI that only answers is still a tool you operate by hand. The loop is what turns it into something that runs the work itself: read, act, repeat, on a trigger or a schedule, inside limits you set. The long-form companion to Episode 6, the finale of Claude Code for Founders: why a smart assistant that waits for you is still a bottleneck, what the loop actually is, what it takes to close it, how you stay in control, and how voice, memory, skills, and live data finally come together into one self-running cycle.