Use cases · Multiplayer session
Two people. One shared conversation.
An AI that knows when to speak.
A live three-way session: Alice teaches Bob his first programming lesson in a shared Senua AI conversation. Senua generates the code examples when it’s asked to — and stays silent while the two of them talk to each other. No copy-pasting between windows, no bot butting into every message.
The scenario
A first programming lesson, taught live.
Alice is teaching Bob to code — a classic first lesson: what a program is, and the traditional first line every programmer writes. Instead of a screen-share and a separate chatbot window, they work in one shared Senua AI conversation. Alice explains the concepts in her own words; when an example is needed, either of them asks Senua directly and the code appears in the shared thread for both to see.
The quiet star of the session is what Senua doesn’t do. Most assistants answer every message — every aside, every “good work”, every human-to-human sentence gets a reply. Here, when Alice is talking to Bob and nothing is being asked of the machine, the machine says nothing. The conversation stays a conversation.
What the session shows
All of it live, private, and on ordinary hardware — the same single mind serving both participants.
1 · A genuinely shared session
Alice shares a project; Bob joins it live. Both sides chat in the same conversation from their own accounts and machines — messages, and Senua’s replies, appear on both screens, each correctly labelled from that person’s point of view.
2 · Code on request
“Senua, write a python program that prints Hello, world.” The example arrives as a clean, copyable code block in the shared thread. The student can ask for his own variation and get his own example — both participants are first-class.
3 · The judgement to stay silent
Teacher-to-student explanations pass with no interruption. Senua answers when it’s being asked and holds back when it isn’t — and when it has nothing grounded to offer, it stays out rather than guessing.
The lesson plan
The session, as scripted.
Five interactions — three between the humans, two to Senua. This is the plan the live recording follows; the labels show who each message is for, and what Senua does about it.
Alice → Bob
“Welcome Bob! First lesson: a program is just a list of instructions the computer follows, one line at a time. We’ll start by making the computer print a message.”
Senua stays silent — two people talking, nothing asked of it.
Alice → Senua
“Senua, write a python program that prints Hello, world.”
Senua → the shared thread
A generated, copyable example — visible on both screens:
print("Hello, world")Alice → Bob
“Look at the example above, Bob. The first line tells the computer what to show on the screen, and the words in quotes are what it shows. Change those words and the message changes.”
Senua stays silent again — the explanation is Alice’s to give.
Bob → Senua
“Senua, write a python function that checks if a number is odd.”
Senua → the shared thread
The student’s own request, answered in the same shared conversation:
def is_odd(n):
return n % 2 == 1Alice → Bob
“Homework: read both snippets and tell me next time what def and return do. Good first session, Bob.”
Senua stays silent — the lesson ends between the people it belongs to.
Why it’s hard — and why this matters
Collaboration, not a chatbot with an audience.
One conversation, many people.
Both participants are signed in as themselves, on their own machines, in the same live thread. Each screen shows “You” for its own user and real names for everyone else — the same session, correctly told from two points of view.
Knowing when not to answer.
Group conversation is where assistants embarrass themselves — replying to messages that were never for them. Senua AI decides, message by message, whether anything is actually being asked of it; human-to-human turns pass without interruption, and when it has nothing grounded to say, it says nothing rather than guessing.
A third participant, not a tool in a tab.
The AI shares the room: it sees the same thread the people do and contributes on request. No copy-pasting answers between a chatbot window and the group chat.
Private, and on ordinary hardware.
The whole session — both users, the shared project, the generated code — runs on one everyday machine. No GPU, no cloud, no per-seat AI licence.
Put a mind in the room with your team.
Shared sessions, live answers, and the judgement to stay out of the way — on your hardware, with your data. This is one of an ongoing series of end-to-end use cases.