The blinking cursor.
You opened Claude Code or Cursor. It can do anything. So you typed something vague and got something vague, closed the terminal, and went back to ChatGPT.
54 files · one-timeBuilt with Claude Code and Codex
The business system your AI can run.
13 agent-ready modules, in order, with a business brain your agent loads every time and trackers that hold state. Drop the folder into Claude Code, Cursor, or Codex, send one message, and Day 1 exists by tonight.
GOStart Here: the 3-day setup
1Day 1: business brain + profit leak audit
2Day 2: content engine + tracker
3Day 3: first automation + first decision
+Then: launch, ads, client work, agent board, 30-day rhythm
Stripe checkout. Zip download in minutes. Lifetime access. 3-day setup guarantee.
Claude Code
Cursor
Codex
OpenClawOr plain Claude and ChatGPT: open a module, paste the prompts. Same result, more clicks.
This shop, the ads that sent you here, and the emails that follow were built with these files in Claude Code and Codex.
In your words
You have the engine. You do not have the road.
The prompt doc.
There is a Notion page with forty saved prompts. You have not opened it since the week you made it.
The tracker.
You built a content calendar once. It has three rows. It stopped being true a month ago.
None of that is a tools problem or an intelligence problem. It is a sequence problem. Nobody told you what to build first, what inputs to gather, what done looks like, or where the result should live so it still exists tomorrow.
The kit is that sequence, written down, in files your agent can run.
The mechanism
Written for your agent, not just for you.
Every module has the same shape: what it produces, the inputs it needs, the complete prompts, an agent build brief that names the exact files to create, a definition of done, and a worked example. AGENTS.md at the root tells Claude Code, Cursor, or Codex how the whole kit runs: the order, the rules, where to save state.
Business brain
Module 05 interviews you and writes scalr/business-brain.md. Every later module reads it. Teach it once.
Build briefs
Each module tells the agent what to make and where to put it. Drafts, not chat.
State
Everything lands in a scalr/ folder in your project: audit, backlog, decisions, batches, trackers. Tomorrow starts where today ended.
Definition of done
A checklist in every module. The agent reports it. You know when it is finished.
Install it with one message
First message in Claude Code
Read AGENTS.md in the ai-scalr-kit folder, then run Day 1 of START-HERE.md. Ask me only for the inputs you are missing.
What your agent builds on Day 1:
scalr/business-brain.md
scalr/audit-2026-09-03.md
scalr/automation-backlog.md
No agent? Every module works pasted into plain Claude or ChatGPT. Module 13 has the fallback.
Day 1 by tonight
The 3-day setup.
Day 1
60 to 90 minutes
Your agent interviews you and writes the business brain. Then it runs the Profit Leak Audit: seven leak types, a scorecard, a ranked fix list. You end with three files and one leak you are fixing this week.
Day 2
60 minutes
The content engine. One idea in, a 14-day calendar out, with 10 hooks, 3 scripts, 3 carousel outlines, and a tracker that is now yours. The quality checklist runs on the first three.
Day 3
60 minutes
The Automation Opportunity Finder scores your ten most repeated tasks and writes the first automation out in full. Then you bring one decision you have been avoiding to the AI Boardroom and log the answer.
After Day 3 the system exists. Everything else in the kit is expansion.
Lifetime access. No subscription. No upsell.
What is inside
13 modules. 4 libraries. 3 command centers. 54 files.
01
AI Content Engine
One idea becomes hooks, scripts, carousel outlines, captions, and a 14-day calendar. Agent builds scalr/content/ and your tracker.
02
100 Hooks and Scripts
50 hooks, 10 script frameworks, three worked niche adaptations, a scoring rubric, and a hook-to-script-in-5-minutes prompt.
03
Profit Leak Business Audit
Seven leak types, scorecard with scoring guidance, a pain-by-frequency-by-ease matrix, a 20-minute fix list. Agent writes the audit and the backlog.
04
AI Boardroom Advisors
Content, offer, growth, and operations advisors with full role prompts, a boardroom meeting that forces one decision, a decision log.
05
Personal AI Agent Setup
The business brain: 25+ fields with hints, the agent system prompt, a drop-in CLAUDE.md snippet, a memory update prompt, a weekly refresh.
06
Automation Opportunity Finder
Inventory, five-factor scoring, the build brief template, and three complete no-code automations written out end to end.
07
Launch Stack
Offer builder, page sections with copy prompts, lead magnet, checkout bullets, 5-email sequence, launch calendar. Ten supporting files and a Launch Command Center.
08
Higgsfield Ad Lab
Concept inputs, UGC prompt, scene prompt, scorecard, testing loop, agent dashboard path. 100 prompts and a Creative Testing Command Center behind it.
09
Client Automation Service Templates
Audit, offer, proposal, scope, timeline, handoff, and how to think about pricing, for selling AI systems to clients. Eight supporting files.
10
Command Centers and Trackers
How every dashboard and CSV works, the weekly review, and a brief that builds your own scalr/dashboard.html.
11
30-Day AI Success Plan
The rhythm after setup, with what done looks like each week and a monthly review.
12
Agent Operations Board
Roles, tickets, governance rules, a weekly board, and a run-my-board prompt for running several AI helpers with approvals.
13
Agentic Tool Execution Guide
Install and run the kit in Claude Code, Cursor, Codex, OpenClaw, or plain chat. Troubleshooting included.
content-engine/
9 files50 hook templates, 50 script frameworks, generator prompts, 10 example scripts and captions, a niche example bank, the quality checklist, a content tracker, a 30-day calendar, the Content Command Center.
launch-stack/
10 filesOffer worksheet, website copy templates, checkout setup guide, lead magnet page, email capture copy, 5-email sequence, funnel map, landing page prompt pack, 30-day launch calendar, the Launch Command Center.
ad-lab/
9 files100 Higgsfield ad prompts, angles, templates, testing SOP, testing spreadsheet, script-to-video workflow, agent dashboard blueprint, bonus mini-pack, the Creative Testing Command Center.
client-services/
8 filesKickoff questionnaire, offer structure, scope, timeline, handoff checklist, landing page buildout checklist, lead magnet template, checkout and email flow map.
ai-scalr-kit/ START-HERE.md AGENTS.md 00-kit-map-and-3-day-setup.md README.md modules/ 13 files content-engine/ 9 files launch-stack/ 10 files ad-lab/ 9 files client-services/ 8 files
Read one before you buy
This is module 03 as it ships. Every module has this shape.
# 03. Profit Leak Business Audit A 45-minute audit that finds where your business loses time, follow-up, and clarity, scores it, and hands your agent a ranked backlog of fixes, for any owner who suspects the leak but cannot name it. ## Use this when - You are busy all day and cannot say which hour paid for itself. - Inquiries come in and some never get an answer. - You rewrite the same email, quote, or reply every week. - You are about to buy a tool or hire someone and want to know what the actual problem is first. - Day 1 of the setup: your business brain is filled and you want the first honest look at operations. ## What you walk away with - `scalr/audit-YYYY-MM-DD.md`: intake, seven leaks with evidence, a scorecard, a prioritization matrix, and a 20-minute-fix list. - `scalr/automation-backlog.md`: everything worth templating or automating, ranked, with the first three marked. - One fix done today. ## Time First run: about 45 minutes. Rerun: about 20 minutes, monthly. ## Inputs ```text Business: what you do and for whom, one line Offer and price band: the main thing you sell. A band is fine if you would rather not paste exact prices. How people find you: referrals, search, social, walk-in, a directory, paid How an inquiry arrives: phone, form, DM, email, front desk. Who answers, and how fast on a normal day. Sales process: first contact to paid, the real steps, including the messy ones Follow-up process: what happens when someone does not book or reply. "Nothing" is a valid answer. Content process: what you publish, how often, who does it. "Nothing" is valid. Onboarding: what a new customer gets in the first week Weekly repeated tasks: every task you or staff repeat weekly, with rough minutes each Tools in use: the software you pay for, and what each is actually used for Team: who does what. Solo is fine. Where you feel stuck: one sentence, plain words Last three things that went wrong: real incidents. No names needed. ``` ## The system ### Step 1. Intake check
See a run
What Day 1 looks like.
Before
my-business/ (empty)
After Day 1
my-business/
ai-scalr-kit/
scalr/
README.md
business-brain.md
audit-2026-09-03.md
automation-backlog.md
decisions.mdDrafts, not decisions. You review everything before it goes anywhere. The agent cannot publish, send, buy, or deploy. That rule is in the files.
What it replaces
If you hired these out.
| What you would hire out | Typical cost | In the kit |
|---|---|---|
| Business audit, consultant half-day | $1,200 | Module 03 |
| Advisory board, four advisors, one month | $1,000 | Module 04 |
| AI setup and business brain, done for you | $1,500 | Module 05, AGENTS.md |
| Automation consulting, map plus three built automations | $1,500 | Module 06 |
| Content strategist, one month | $1,500 | Modules 01, 02, content library |
| Launch copywriter, offer to emails | $2,000 | Module 07, launch library |
| Ad creative lab with testing system | $750 | Module 08, ad library |
| Service packaging, audit to handoff | $900 | Module 09, client library |
| Ops layer: board, trackers, 30-day plan | $650 | Modules 10 to 12 |
| Hired out | about $11,000 | The kit: $37, once |
Typical market rates for the named services. The kit gives you the systems and your agent does the work. Your judgment stays yours.
No subscription. No upsell. No login. A zip and a folder.
Lifetime access. Re-download from your link whenever the files update.
Use it on client work. Keep the deliverables. Do not resell the files.
Plain text. Open every file in any editor.
3-day setup guarantee
Follow Start Here. Within 3 days the kit is running inside your agent, or your chat workflow. If it is not, email or DM Sohum with what you ran and where it stalled, and he works through it with you until it runs.
All sales are final. This is a support promise, not a refund, and for a product that is files, it is the one that matters.
Who it is for
For the stuck builder. Not for everyone.
For you if
- You have Claude Code, Cursor, Codex, or ChatGPT and no system around it.
- You run a creator business, a service business, a freelance practice, or a project you want to become one.
- You want the order of operations, not more theory.
- You would rather have three files that exist than forty prompts you saved.
Not for you if
- You want one prompt to paste and stop there.
- You have no business, offer, or project yet. Start with the free Starter Routine.
- You want promises about income, views, or leads. There are none here.
- You are an engineer who will build your own operating layer. You probably will.
Built with the same files
A note from Sohum
I run Startscalr, a marketing systems company. The shop you are reading, the ads that brought you here, the email campaign that follows, and the product files themselves were built in Claude Code and Codex, with the business brain and the modules in this kit doing the sequencing. I did not write a course about a workflow. I packaged the one I use, and I answer DMs when something in it does not run.
Clients Sohum built systems for, using the same workflows.
Nishat Lis
Sohum helped me set up a real content workflow from scratch. Structure instead of random prompts.
Mohammed Ali
Real estate investment group
Sohum helped me build a website and automations that made us look more professional and follow up faster.
Ljudmila Brun
Sohum helped me audit my business and find where I was wasting time. The advisor prompts felt like having a strategist in my pocket.
Ashwin Shah
Sohum helped me turn one idea into scripts, captions, and carousels in a single session.
Ushi Patel
Beauty and skincare founder
The kit made it easy to plan content, write product descriptions, and run my socials without feeling overwhelmed.
Shyam Kaur
Sohum helped me package AI automation into something my clients could understand.
Larry Kaluža
Every section had a clear next step and it stopped me from spinning my wheels.
FAQ
Questions that stop the sale.
You can get a one-off answer. You cannot get a sequence, a business brain the model reads every time, trackers it updates, or a definition of done. Prompts are telling a stranger your job every morning. The kit is the binder the new hire gets on day one. It makes a good model consistent. It does not make a bad prompt good.
Yes, and CSV and HTML. That is the format agents read. Unzip it and there is no app, no login, nothing to install. What you are paying for is the sequence, the build briefs, the state files, and a person who answers when it stalls.
Claude Code, Cursor, Codex, and OpenClaw, tested. Any agent that can read a folder. And plain Claude or ChatGPT: open a module, paste the prompt blocks, save the outputs where it tells you. Module 13 covers each path.
No. You unzip a folder, open it in your tool, and send one message. The agent asks you questions. If you have never used a terminal, module 13 has the plain-chat path.
Day 1 is 60 to 90 minutes and ends with your business brain, your audit, and a backlog, saved as files. The first module runs in about 15 minutes once the brain exists.
It means the kit is running inside your agent within 3 days, or Sohum works through it with you until it does. Email or DM with what you ran and where it stalled. All sales are final, and this is the promise that matters for a product made of files.
No. Nothing in the kit promises outcomes, and the prompts are written so the model will not either. It helps you build systems faster and with fewer decisions. Results depend on your offer, audience, and follow-through.
Yes. Your access link does not expire. When the files update, re-download the zip from the same link.
No. One product, one price, no subscription. The free gifts on the homepage are free.
Yes. Use the modules and the client-services library to audit, propose, scope, and deliver for your clients, and keep the deliverables. Do not resell, redistribute, or give away the kit files.
It happens when files are long and vague. The kit's files are short and specific, each module is self-contained, and AGENTS.md is written as rules, not prose. If it still drifts, module 13 has the fix: run one module at a time and name the file.
A Claude, ChatGPT, Cursor, or Codex plan, whichever you already use. Nothing else is required. Some workflows are nicer with a design tool or a spreadsheet you already have.
Day 1 exists by tonight.
Unzip the folder. Send one message. Answer the questions. By the end of the evening you have a business brain your agent reads every time, an audit that names your first fix, and a folder that will still be there tomorrow.
Get the kit, $37Lifetime access. 3-day setup guarantee. Stripe checkout.
Questions before you buy? DM @startscalr_sohum. Sohum answers.
AI SCALR KIT
$37 once