If you also have five chats open with your AI agent, the software that does work for you on the computer, and you carry text between them by hand, you're not alone.

Where you are now

Could it be that the chats aren't the problem? There are three problems here: you are the messenger, every chat checks itself, and nobody sees the whole picture except you.

Every piece of information passes through you, and every copy-paste is one more place for a mistake. And a chat's “done” is a claim, not proof: you open the website, and the old page is still there.

And which chat stopped halfway an hour ago? You don't know, because nobody told you.

Axel Jedlitschka

Who's talking to you: Axel Jedlitschka

It all started with an income and expenses table I wanted to build for myself in Excel. After a few versions I moved to Google Sheets, and from there it kept growing.

While booking sales meetings by phone, I started building systems and automations, processes that run on their own, for business owners. I built a system for business forecast analysis, and it served the company's marketing and sales very well. In a system with several parts, one person has to hold the whole picture, and check each part before the next one starts.

Axel Jedlitschka

What to demand

  • Instructions ready to paste

    Get each chat its complete instructions, and paste them as they are, with nothing to assemble and no “replace line four”.

  • The cost before you choose

    Know what each option costs before you pick it: which model, meaning which version of the AI, how much effort, and how many chats.

  • A “done” that was checked

    Know that every “done” was checked with a measurement, not taken on the chat's word.

  • A stuck chat gets caught

    Know when a chat stopped halfway, even when nobody told you.

  • The next task after a check

    Know that the next task goes out only after the previous one was checked.

There is a way

One chat gets one job: orchestrator. It looks at where the project stands, decides what to do and on which model, writes every other chat a complete prompt, meaning the full instructions you paste into it, and checks what comes back. It does not do the work itself, so it stays cheap, and it checks work it did not do.

  1. Map.At the start of every round the orchestrator checks what actually exists: the instruction files, meaning the documents that tell the agent how to work, each chat's folder, and the reports that came back. It does not assume; it looks.
  2. Decide.Every option it offers comes with a model, an effort level and a number of chats, so you know the cost before you choose.
  3. Full prompt.For each chat it writes a complete prompt from a template: the model and effort level on the first line, then the context, the task, what to hand off, stop conditions and a one-line report. You paste; you do not assemble.
  4. Verify.When a chat reports, the orchestrator measures for itself: it runs the check, requests the page, sends a test lead and confirms it arrived. Only then does the next task go out.

A chat's “done” is a claim, not proof.

The way, in one kit

Orchestrator kit · a kit by Axel Jedlitschka

One chat runs all the others: it hands out the work, and checks every result itself.

  • At the end of the install it checks its own work, and if something failed it puts the previous version back
  • It shows what it is going to do, and asks before it writes anything
  • The method works in any AI tool that runs several chats, such as Codex, Cursor or Gemini CLI
  • A Claude Code skill called orchestrator: a file of instructions that turns one chat into the orchestrator
  • A prompt template for each chat, with the model and effort level on the first line, in English, Hebrew and Spanish
  • A template for the work map: for each chat a folder, what it reads, where it hands off and when it is done
  • A verification guide, a worked example, and a 22-check checklist
  • A small program that spots a chat that got stuck halfway
  • A block of instructions ready to paste into Codex, Cursor and Gemini CLI
  • Install in one command, with a backup of any earlier version
The first page of the guide: One orchestrator instead of five chats

The comparison

Free

Orchestrator kit

Who carries the information
The orchestrator, in a complete prompt for each chat
Who checks the “done”
The orchestrator measures for itself, before the next task
Who sees the whole picture
The orchestrator, and you get one line

A chat for each part

Who carries the information
You, copying from chat to chat
Who checks the “done”
Every chat checks itself
Who sees the whole picture
Nobody except you

The offer

A skill, a prompt template, a work map, a 22-check checklist, and a guide of 7 pages

Price: FreeNo credit card, and no part is held back for payment

The kit is free and you get it right away. You only get marketing messages if you tick the box, and you can unsubscribe at any time.

As soon as you leave your details, a thank-you page opens with a button to download the kit. You choose the guide's language in the form: Hebrew, English or Spanish.

By Axel Jedlitschka

What you get

Promised

  • The full kit, free, as soon as you leave your details.
  • No part is held back for payment.
  • Your details are not sold, and are not passed to anyone except the infrastructure providers that run the site.
  • A question about the kit? Send a message on Instagram to @theonlyaxelj.

Not promised

  • A result. The kit teaches the agent to manage and check, and the result depends on how you apply it and on your business. It is general knowledge, not personal advice.
  • That the orchestrator does the work. It plans, writes and checks, and the other chats do the work itself.

There is no purchase and no payment here, so there is nothing to cancel. You can withdraw your consent to marketing messages at any time.

Questions?

How do I get the kit?

You leave your details, and a thank-you page opens right away with a button to download the kit.

Is it really free?

Yes. No credit card, and no part is held back for payment.

I'm not a programmer. Can I install it?

Unzip the file you downloaded, and from the folder that opens, run one command in the terminal, the window where you type commands to your computer: bash install.sh. The install shows what it is going to do and asks before it writes anything. The default answer is “no”.

Does it only work with Claude Code?

The skill is built for Claude Code. The method works in any tool that runs several chats, such as Codex, Cursor tabs or Gemini CLI. For them the kit has the file AGENTS.md and the templates.

Why not one chat that does everything?

Because it bloats: every file, error and output lands in its conversation, and every turn costs more. And whoever did the work is not the best one to check it.

How many chats do I need?

One per kind of work that needs its own context or folder. Often two is enough, and one small task needs no orchestrator at all.

Will I get spam or phone calls?

Your details are used to send you the kit and to follow up on it. Marketing messages are sent only if you ticked the consent box, and you can unsubscribe at any time. You can also ask to see, correct or delete your details.

Where can I get help?

A question about the kit: send a message on Instagram to @theonlyaxelj.

What now

You can keep copying text from chat to chat and checking every “done” alone, or you can let one chat run and check all the others.

A kit to install and a PDF guide of 7 pages: a skill, a prompt template, a work map and a 22-check checklist, free.

Price: FreeNo credit card

The kit is free and you get it right away. You only get marketing messages if you tick the box, and you can unsubscribe at any time.

The kit is ready to download as soon as you leave your details

The kit is free and you get it right away. You only get marketing messages if you tick the box, and you can unsubscribe at any time.

Your details are used to get back to you. What is collected, why, and who receives it: see the Privacy Policy (opens in a new tab).