Academy
1. Meet @oupi — More Than a Chatbot
Understand what @oupi is, how it differs from a simple chat, and the different entry points you can use to reach it.
@oupi is not a chatbot — it is the agent behind the entire OUPI platform. When you give it a goal, it does not just reply with text: it builds a plan, executes it step by step using the right tools (search, documents, data, integrations, browser, code sandbox), verifies the outcome against your original goal, and delivers the result. Every unit of work is called a mission. A mission has a plan, a cost in credits, a verification verdict, and deliverables. Think of it as delegating a task to a capable assistant who asks smart questions, does the legwork, double-checks the output, and hands you the finished product.
You can reach @oupi from several entry points, but it is always the same agent underneath:
• "Talk to Oupi" in the header — quick, conversational access; you can launch a mission from here. • The Assistant — @oupi's cockpit, where you see today's agenda, all your missions, items waiting for your decision, wake-ups (scheduled tasks and watchers), and settings. • Studio & CodeBuilder — @oupi produces documents, spreadsheets, presentations, or applications on your behalf. • ChatPro — for deeper conversational work.
No matter where you start, the mission follows the same lifecycle: goal → plan → execution → verification → delivery.
Before @oupi starts working, it may ask you a clarification — a single question or a short series — when something essential is missing or ambiguous. It prefers asking over guessing. Your answers are saved with the mission so nothing is lost. Once the work is done, every mission ends with a verification verdict: verified, failed, or could not be verified. That last one is not a failure — it simply means the checker could not read the output (e.g., a binary file or an external action) and hands it to you for review. You'll find all items needing your input in the "To decide" list, where you can approve, reject, snooze, or cancel.
Write missions as outcomes, not step-by-step instructions. State what you want, for whom, in what form, and any hard constraints. For example: "Prepare a one-page briefing on Q1 sales for the board, as a PDF, using only internal data." @oupi decides the steps — you define the destination.
@oupi can work even when you are not there. Wake-ups let you schedule recurring tasks ("every Monday at 8 AM, compile a press review") or set watchers that react to events (a new email matching a rule, a change in a connected tool). Each wake-up can be paused, armed, tried once, and keeps a full history of its runs. Deliverables from scheduled missions land in a durable Studio folder and can also be emailed to you automatically.
Open the Assistant now (navigate to the Assistant cockpit). Explore the four sections: Today, Missions, To decide, and Wake-ups. Then type a simple goal in "Talk to Oupi" — for example, "Summarize the key features of @oupi in three bullet points" — and watch the mission lifecycle unfold: plan, execution, verification, delivery.
@oupi is an agent, not a chatbot: it takes a goal, plans, executes with tools, verifies, and delivers. Every task is a mission with a cost, a verdict, and deliverables stored in Studio. You reach it from the header, the Assistant, Studio, or ChatPro — same agent, different doors. It asks clarifications when needed, and the "To decide" list keeps you in control. For hands-off work, set up wake-ups. Write clear outcomes, and let @oupi figure out the steps.