The easiest way to assess ChatGPT Intelligent UI is to give it a small task whose answer you can check. Choose work that currently involves rearranging information, comparing options or asking several “what if” questions.
The four examples below are suggested experiments, not tested demonstrations or guaranteed features. Use fictional or non-sensitive material for the first attempt. Keep workplace information within your organisation's approved arrangements.
Check your setup first
OpenAI's current model guide says Intelligent UI works with GPT-6 in Chat at Instant through Extra High, where available. Pro effort uses Astra and does not support it; a Pro subscription is a separate matter. Use the web or a supported updated app. Older macOS and Windows ChatGPT apps are unsupported.
You do not need a special activation prompt. Describe the task and ask for the presentation that would help you.
1 Compare options against the same criteria
Start with three fictional venues for a team workshop. Give each one the same fields, including a few deliberately missing details.
Try: “Compare these workshop venues side by side. Let me focus on capacity, travel time and equipment. Mark missing information as unknown and explain which details I need before choosing.”
Check whether the comparison keeps the original facts intact. If an unknown becomes a confident claim, ask for a correction before continuing. The useful outcome is a shorter list of questions to resolve, rather than a recommendation that conceals the gaps.
2 Explore a workload assumption
Use a simple planning exercise with invented numbers. For example, six tasks might each need two hours of preparation and one hour of review.
Try: “Create an interactive workload calculator using these task counts and hours. Let me change the number of tasks and review time. Show the formula and total hours. Keep preparation and review separate.”
Calculate one case yourself. With six tasks at three hours each, the total should be 18 hours. Then try zero tasks and a longer review time. Watch for rounding, hidden assumptions and totals that stop updating.
This is a calculation exercise. Real staffing decisions need wider context, including availability and the nature of the work.
3 Make a process easier to explain
Pick a process you already understand, such as how your team moves a draft article through review.
Try: “Turn this process into an interactive explanation. Show what enters each stage, who reviews it and what is required before it moves on. Use only the steps I have supplied.”
A useful result should help a new colleague ask sensible questions. Check for invented stages and unclear handovers. Ask for a plain-language summary alongside the visual so people can follow the process without relying on the interaction.
4 Plan a focused working session
A workshop agenda is another manageable experiment.
Try: “Help me plan a 90-minute strategy session. Keep 15 minutes for agreeing next steps. Let me adjust the time for each discussion and flag when the total exceeds 90 minutes.”
Test the total and see whether changing one section affects the rest as expected. Then inspect the agenda itself. A mathematically tidy schedule can still ask people to solve too much in one meeting.
Keep the experiment small enough to learn from
After trying one example, note what became easier and what needed correction. Include the time spent checking the output when judging whether it helped.
Repeat the task with a simpler presentation if necessary. The aim is a working method you would willingly use again.
For help shaping a client-owned dot around recurring professional work, explore AGENTICA IX Agents.