Presentation Designer is an Agent Exchange agent in Autonomous Insights. It uses your survey data and a PowerPoint template to generate a stakeholder-ready deck in minutes. This article outlines supported file types, template requirements, data structure, and workflow.
Presentation Designer runs as a single, continuous workflow. Plan to complete the process in one sitting: upload files, define objectives, map questions, and generate the deck.
Save-in-progress is not available.
Access the Presentation Designer through the Agent Exchange in your Research Tools.
- In the Fuel Cycle platform, click the Research Tools button in your left navigation menu.
- Click Autonomous Insights.
- Click Agent Exchange.
- Click Presentation Designer.
Presentation Designer does not run in the core Fuel Cycle user experience.
File Details
PowerPoint Template
Each uploaded PowerPoint file must be 200 MB or smaller. Supported formats are .ppt and .pptx.
There is no published hard limit for the data file, but performance improves when both files are focused and lightweight.
Large templates with high-resolution images or unused slides can slow down generation. Oversized datasets that include multiple waves, regions, or segments can also make mapping harder than it needs to be.
If a file feels heavy to open manually, trim it. Remove unused slides from your template and limit your dataset to the survey and variables required for this presentation.
Use a standard .pptx file. If your deck is in Google Slides, Keynote, Portable Document Format, or an older PowerPoint format, export it as a .pptx file before uploading.
Your template must use slide masters and layouts. Presentation Designer relies on real PowerPoint placeholders to insert titles, charts, and body content. If key text is embedded in images or scattered across manually placed text boxes, the agent cannot correctly place the content.
You do not need a large layout library. A small, consistent set works best, such as a title slide, a title-and-content layout, and one or two charts.
Template Best Practices
Presentation Designer builds directly on your template. Clean structure leads to better output.
Use slide masters and structured layouts. Keep titles and body content inside actual placeholders. Avoid embedding section headers or key text in background graphics.
Keep layouts predictable. If an analyst can quickly add new findings to your template without hunting for the right text box, the agent will perform well. Consistency matters more than variety.
Data File
Upload a .csv or .xlsx file with a flat, tabular structure. Each row should represent one respondent. Each column should represent one question or variable. The first row must contain column headers.
Most exports from Fuel Cycle Surveys or third-party survey tools work with little or no cleanup as long as you save them as CSV or Excel files.
Data Structure and Cleanup
Your data file should be simple and easy to read.
Use clear column names such as Q1_OverallSatisfaction or BrandPreference.
Survey questions are the core of presentation creation, and most survey files are filled with a ton of noise in the form of demographics. So, by default, we filter out demographics. If you have a demographic question you want included in the analysis, you have to reword it for AI to identify it as a survey question.
- Avoid vague labels like Q1, col_A, or VAR_001.
- Descriptive headers make mapping faster and reduce confusion.
- Before uploading, remove empty columns, and confirm that scales and coding are consistent.
- Ensure demographic variables such as age, gender, and region are in separate columns.
- Avoid merged cells, embedded charts, or pivoted layouts.
Presentation Designer does not correct structural issues in your dataset, so small cleanup steps make a big difference.
Objectives and Question Mapping
After you upload your template and data, you will name the presentation and define its objectives. Objectives describe what the deck should show, such as summarizing overall satisfaction by segment or comparing brand awareness and consideration.
Next, you map questions from your dataset to those objectives. One question can support multiple objectives. Clear objectives and descriptive column names make this step straightforward and improve the final output.
Once mapping is complete, run the analysis and generate the deck.
Troubleshooting
If the generated deck does not look right, review your files first.
Open your template and confirm that titles and content use real placeholders, not images or scattered text boxes. If the slide masters are cluttered, simplify them.
Then review your data file. Confirm that it contains one flat table with a header row, one row per respondent, and clear column names. Remove unrelated surveys, empty columns, or unnecessary complexity.
If needed, test with a simplified setup: a streamlined template and a smaller dataset with only the variables you plan to present. Once that works, you can scale up.