AI gives generic answers because it knows nothing about your business.
Ask an AI to write to a potential client and you get a polite, empty message. It could be from anyone, to anyone. The model is not the problem. It has never seen your offer, your best clients or the reason your last proposal worked.
Admin hits the same wall. Take a question like “Where are the photos from the March site visit, and who is still waiting for a reply?” The answer is spread across your inbox, a downloads folder and someone’s phone. An AI cannot find what it has never been shown.
Most people deal with this by retyping background into every new chat. That works for a day. Then it becomes a chore, and the tool slides back into being something you try now and then.
The fix that lasts is to write your context down once, in files you own. Then organise it so the AI reads the right part at the right moment. FullDigital teaches this as five layers. You still choose the model. The context stays yours.
Try this before your next client message
Paste three lines above your request:
- Who you help and what problem you solve, in one sentence.
- One real detail about this person or company.
- One rule, for example: “Keep it under 100 words and end with one question.”
Then compare the reply with what you got without those lines. The five layers make that difference permanent.
Five layers turn a chat window into a working system.
Each masterclass starts here, so beginners can follow. These are the same layers FullDigital explains on its AIOS pages. Each example below comes from one of the two masterclass topics: finding clients, or keeping email and files in order. They are illustrations, not templates you must copy.
- Layer 1
Constitution: the rules and standards
This is the first thing your AI reads, every session. It says how you work and what the AI must never do without asking.
Example: “Never send anything to a client without my approval. Write in plain English. Ask before you assume a price or a date.”
Why it helps: You stop repeating your standards, and the AI stops guessing them.
- Layer 2
Memory: the context you keep
This is what your AI should know about your business. It lives in files you can open and correct yourself.
Example: One short file per client covering who they are, what they bought, what they asked last time and how they like to be contacted.
Why it helps: A message to a returning client starts from what actually happened, not from a blank page.
- Layer 3
Skills: repeatable ways of working
A job you do often, written down once as steps, so it gets done the same way each time.
Example: A “first contact” skill. Read the client file. Pick one relevant detail. Draft three sentences. End with one clear question. Show me the draft before anything is sent.
Why it helps: Good work becomes repeatable. When something goes wrong, you fix one step instead of explaining the whole job again.
- Layer 4
Tools: services the AI can use
These are the connections to where your work actually lives: email, calendar and files. Connectors built on MCP (Model Context Protocol) are one common way to make them.
Example: Let the AI read one project folder and search your inbox, but not send, move or delete anything. Add write access only where you really need it.
Why it helps: The AI works with real information, and you decide exactly what it can touch.
- Layer 5
Focus: what matters now
One short page that says what you are working on this week.
Example: “Reply to the three people who asked for a quote last week. Finish the overview for the office move before Friday.”
Why it helps: The AI starts with what matters to you this week, not with whatever it happens to be asked.
You do not need all five on day one. The free foundations walk you through a first evening: one folder, one constitution file and two skills. The masterclasses show the layers working together on a real task.
Terms you will hear in the sessions.
New to some of these? Each one has a free FullDigital explainer. In the masterclasses they show up as parts of one system, not as separate topics.
- AI (artificial intelligence)
- Software that learns patterns from data instead of following hand-written rules. Chat assistants are the everyday kind most people meet first.
- LLM (large language model)
- The model behind chat assistants such as ChatGPT, Claude and Gemini. It works only with what is in front of it, which is why your context matters.
- Prompt
- What you put in front of the model before you ask. With the five layers in place, most of it is already written, so you do not start from one line.
- AI agent
- An AI that takes steps with tools instead of only answering. The method keeps a person approving anything that is sent or changed.
- MCP (Model Context Protocol)
- An open standard for connecting AI assistants to outside tools and data, such as email, calendars and files. It belongs to the Tools layer.
- Second brain
- Your notes and files organised so you can find and reuse them. A Business Brain is a second brain that an AI can read as well.
Two sessions. Two everyday jobs. Built live on screen.
Both sessions are 90 minutes, live online and in English. Each one starts with the five-layer foundation so beginners can follow, builds one practical workflow live on screen, and leaves you with materials to keep. Join one or both, since they cover different topics. “Business Brain” is the organiser’s name for an AI operating system built around one business: the same five layers, filled with that business’s own context.
17:00–18:30 Riga · 16:00–17:30 CET · 11:00–12:30 New York (EDT)
Client acquisition with a Business Brain
Use your business context to improve how you attract and approach clients. You will see how your own business information shapes the work, instead of a generic sales message.
What you leave with: A usable piece of client acquisition work, and a repeatable way to continue it with your own business information.
This session has taken place.
17:00–18:30 Riga · 16:00–17:30 CET · 10:00–11:30 New York (EST)
Less owner overload: organise email and files with AI
Turn scattered documents, photos and email into a clear working overview.
What you leave with: A linked project overview with files, image previews, outstanding replies and next actions. Then a test of whether the right information can be found again.
This session has taken place.
The New York start time changes between the two dates because the US moves its clocks back on 1 November, a week after Europe.
Made for people who run the work, not for AI specialists.
This is for you if
- You run a small business, work for yourself or lead a team. You already use ChatGPT, Claude or Gemini, but every chat starts from zero.
- Finding the right email, file or photo takes longer than the task itself.
- You want to understand why a method works, not just copy a prompt.
- You are new to this. Each session starts with the foundations.
This is not for you if
- You want it done for you. This is a class: you watch the work being built and keep the materials to carry on yourself.
- You need the session in another language. Both are in English.
- You want an AI that acts with no one checking it. The method keeps a person in charge of what the AI may read, write and send.
Who organises this, and why it is on a nonprofit site.
FullDigital is a Latvian nonprofit association (Reg. no. 40008303188, Ikšķile, Latvia). We teach AI literacy for free through open resources, public talks and projects that help people build their own AI operating system. This page is part of that work. The explanations on it are free to read and need no signup.
The two masterclasses are organised by Arete AIOS, a company. They are hosted by Egils Boitmanis, founder of Arete AIOS, who also chairs the FullDigital board, and Gianluca Giuliani of Living Off AI.
The masterclasses are free. You do not need to buy anything to attend or to keep the materials. Arete AIOS also offers an optional paid four-week cohort for people who want guided practice on their own business. It is separate from the masterclasses, and you can ignore it. FullDigital receives no payment for this page or for registrations.
We link to these sessions because they teach the same five-layer method FullDigital explains openly on this site. Commercial AI work for companies, beyond what a nonprofit can responsibly do for free, happens separately at areteaios.com.
Questions people ask before they register.
Is it really free?
Yes. Both masterclasses are free to attend, and you do not need to buy anything. The organiser, Arete AIOS, also offers an optional paid four-week cohort. You do not need it to join, or to use what you learn.
I am new to AI. Will I keep up?
Yes. Each session starts with the five-layer foundation, in plain language, before any building begins. For a head start, read the five layers above or the free foundations.
Do I have to join both?
No. The sessions cover different topics: one is about finding clients, the other about email and files. Join one or both.
What time is it where I live?
Both start at 17:00 Riga time, which is 16:00 in Central Europe. In New York the first session starts at 11:00 and the second at 10:00, because the US changes its clocks on 1 November. You get calendar links when you register, and your calendar app shows your local time.
Do I need to share my own files or connect my email?
No. The organiser runs the demonstrations with prepared examples, so you do not need to share personal files or connect your inbox to attend.
What do I keep afterwards?
The session materials, and a free account with the full Business Brain build guide. Everything on this page and on the AIOS pages also stays free to read.
Why does a nonprofit site link to a company?
FullDigital teaches the five-layer method for free. These masterclasses teach the same method live, and they are free too. We say plainly who organises them, so you can decide for yourself.
New dates will be announced.
Start with the free foundations or join the waiting list to hear about the next sessions.