↓
Overview1 · OSWork2 · Jarvis3 · Queues4 · Social media
Sequence 2 of 4 · Assistant

Create your Jarvis

A personal assistant that talks with you on Telegram, remembers your conversations, and has the personality you choose.

Sequence 2 — Create your Jarvis with INTELECTO

INTELECTO is a personal assistant written from scratch in Python, kept small on purpose so you can read, understand, and change it. It receives your messages on Telegram, thinks in an agent loop, searches memory, uses tools, and replies with a cloud model (OpenRouter) or one on your computer (Ollama). Setup takes one command and a six-step configuration assistant. It comes after OSWork because it uses the terminal, folders, and a key stored in the right place.

The video

Sequence 2 — Create your Jarvis with INTELECTO

Video in productionThis page's video is being recorded with Nei's avatar. The video is in Portuguese, with subtitles.
What it is

Small on purpose

No heavy framework, no Docker, no web dashboard.

About 3,000 lines

Python code you can read in an afternoon. You'll understand what it does.

7 dependencies

Fewer pieces to install and almost nothing to break.

Truly yours

It runs on your computer. You no longer have to open a website to chat.

How it thinks

Telegram, agent, model

It's the same design as big assistants, only it's yours.

You on TelegramINTELECTO receives itAgent loop: memory and toolsModel generates a replyReply on Telegram

Cloud model (OpenRouter)

  • Large models, better answers.
  • Pay as you go, with your own key.
  • Doesn't depend on your computer's power.

Local model (Ollama)

  • Runs on your machine, with no cost per message.
  • Your data stays on your computer.
  • Needs memory and preferably a graphics card.
Install

One command and six steps

start.sh sets up the environment and opens the configuration assistant.

git clone https://github.com/inematds/intelecto.git
cd intelecto
./start.sh
1

Create the bot on Telegram

Open @BotFather, send /newbot, choose a name and @username, and copy the token.

2

Paste the token

The assistant checks that the token is valid and shows your bot's name.

3

Choose the model

Cloud through OpenRouter or local through Ollama.

4

Set its name and personality

How it talks: formal, direct, or funny.

5

Send "hi"

Your bot replies with the name and style you chose.

The token and keys stay only in the configuration file, never in text that goes into Git. That's the rule you learned in sequence 1.

Memory

It remembers you

Memory and personality are what make it your Jarvis.

It remembers what you talked about, saves what matters, and speaks the way you chose. Tell it on Monday that you close the sales report every week; on Tuesday, in a new conversation, ask what you do on Mondays. It's not a website that forgets everything when you close the tab.

Practice

Exercises in this sequence

Do them in order. Each exercise has a "ready when" criterion: move on only when it is true.

  1. E1

    Design your Jarvis before installing it · 15 min

    Ready when: the file fits on one screen and someone else would understand who your assistant is.

  2. E2

    The bot on Telegram · 10 min

    Ready when: you have the token and it doesn't appear in any file besides .env.

  3. E3

    Install · 30 min

    Ready when: you send "hi" and it replies with the name and style you chose.

  4. E4

    Memory test · 2 days

    Ready when: it remembers without you repeating yourself.

  5. E5

    Cloud or local · 20 min

    Ready when: you can say which model to use for what, using your own numbers.

  6. E6

    A real routine · 20 min

    Ready when: you used it 5 days in a row and wrote down what to improve.

Open all exercises →

Install it and give it a name

Next, give it real work to do: task queues, in sequence 3.

Open INTELECTO →