Oransim

##Functional characteristics - Causal advertising preview: After entering the advertising copy, budget, talent and platform allocation, predict exposure, clicks, conversion, revenue and ROI. - Counterfactual sand table: Support rapid budget changes and platform proportions on the same launch plan, and observe changes in KPIs. - Multi-layer analysis view: built-in analysis modules such as KPI panel, AI user speech, diffusion curve, world events, comment area debate, and brand memory. - Support by multiple LLM providers: Compatible with OpenAI-style interfaces, and also supports native models such as Anthropic, Gemini, and Qwen. - Synthetic data is available out of the box: By default, it is started with the demo data and world model weights that come with the warehouse, without first preparing a private data source. ###Daily usage 1. When testing a new piece of material every day, directly paste the copy into the Hero on the home page or the 'Material Copy' input box on the left. For example: 'New product of taro milk tea is limited to small coffee shops in summer, new reality try it without stepping on thunder.' 2. Select preset according to your needs first: - `Speed`: It is suitable to see if you can play in one round and quickly get the baseline - `Hotitems`: Suitable for more serious Group chats/comment area debate links - `Complete`: Suitable for more complete full-stack causality analysis 3. If you want the model to actually read copywriting, check the box `Use Real LLM (Let GPT actually read your copywriting)`. At this time, please understand the '15 seconds/90s/ 60- 90s' on the interface as an ideal gear rather than a strong SLA. 4. After clicking `Forecast& Open Sandbox` or `④ View the complete forecast once on the home page, don't just stare at the button copy, and give priority to the left log. As long as it finally appears `is completed...·Sandbox<id>·Baseline ROI...`, And a new record is added to the left side of the `Historical Record ', which means that this round of prediction has truly been completed.🕘 5. After the prediction is completed, first look at the "③ Prediction &amp; Counterfactual Comparison" in the middle and get the six core indicators of 'ROI / CTR / CVR /Exposure/Click/Conversion', and then look at the right side's "④ AI user voting &amp; emotion" to judge "Why did someone order and why did someone skip it?" 6. If you want to continue to optimize this material, don't lose it all over again. Two methods are preferred: - Drag "Douyin %","Little Red Book %" and "Total Budget" directly in the ② Counterfactual sandbox to see the difference between the baseline and counterfactual - On the left side, click "restore to the sand table" and continue to change from an old result🕘 7. If you just want to review old cases, you can directly open History; after refreshing the page or restarting the service, the history of logging in the library will still be preserved.