Academy
3. Handling Clarifications, Approvals and Follow-ups
Respond to clarification requests and approval prompts from the mission page, the Assistant's 'To decide' list, or notifications, and use follow-ups to refine a result without starting over.
When @oupi encounters something essential that is missing or ambiguous in your mission brief, it pauses the mission and asks you — either a single clarification question or a short series. Your answers are saved with the mission and shape how it proceeds. Similarly, depending on the autonomy mode you've set, @oupi may pause to request your approval before executing a sensitive step. In both cases, nothing is sent or published on your behalf while the mission waits for your response.
You have three places to respond to a waiting mission:
- The mission page — open the mission directly and answer inline, right where the question appears in the timeline.
- "To decide" in the Assistant — this is the centralized list of every mission that needs your input: clarifications to answer, results to approve or reject. You can also snooze a decision to come back later, or cancel the mission entirely.
- Notifications — platform notifications alert you when a mission is waiting; tap one to jump straight to the pending question.
Whichever entry point you use, the effect is the same: your answer unblocks the mission and it resumes from where it paused.
When @oupi asks a clarification, answer briefly and directly. If you'd rather it proceed on its own judgment, simply say so — it will make reasonable assumptions and move on. Overly long answers rarely help; a clear, concise reply gets the mission back on track fastest.
Once a mission delivers its result, you don't have to start a new mission from scratch to refine it. Instead, reply directly from the mission page — for example, "shorten it to one page", "add the 2025 figures", or "same analysis but for Germany". This launches a follow-up mission that inherits the original goal, context, and output, so @oupi builds on what it already produced rather than re-doing the work. Follow-ups are the most efficient way to iterate: they cost fewer credits than a fresh mission because the context is already established.
Approvals and clarifications interact with the autonomy mode you've configured. In step-by-step mode, @oupi pauses more often for your sign-off before acting. In fully autonomous mode, it only pauses when something truly essential is missing. Choose the mode that matches your comfort level — you can always adjust it from Studio settings. Remember: every mission also ends with a verification verdict (verified, failed, or could not be verified), so even in autonomous mode you have a final quality gate before using the result.
Open the Assistant and navigate to "To decide". If any missions are waiting, answer a clarification or approve/reject a result. If the list is empty, launch a quick mission from Studio with a deliberately vague brief (e.g., "prepare a report") and watch @oupi pause to ask you a clarification — then answer it from "To decide".
After receiving a deliverable, always check the verification verdict on the mission page before using the result. If it says "could not be verified", it doesn't mean the output failed — it means the automated checker couldn't read it (e.g., a binary file). Review it yourself before sharing.
Missions pause when @oupi needs your input — answer from the mission page, the Assistant's "To decide" list, or a notification. Keep answers brief and direct. To refine a delivered result, use a follow-up reply on the mission page instead of starting over: it preserves context and saves credits. Adjust your autonomy mode to control how often @oupi asks for approval, and always check the verification verdict before using any deliverable.