Once the dashboard already exists, we can also tell the LLM to add new data to the dataset so the visuals stay current without rebuilding the base work. The idea is to reuse the same structure, append new rows, and refresh the dashboard so it reflects the latest information.
Append these new rows to the existing `Retail Analytics Overview` dataset. Do not replace the previous records. Keep exactly the same columns as the original CSV and then update the dashboard so it recalculates KPI cards, charts, and filters.
```csv
year,month,month_number,country,channel,category,product,customer_segment,units_sold,revenue,profit,discount_pct,marketing_spend,website_visits,conversion_rate,avg_order_value,returns,customer_satisfaction,temperature,holiday_season,competitor_price_index
2026,January,1,USA,Ecommerce,Electronics,Wireless Headphones,Returning,402,36120,14045,12,5180,68420,4.12,89.95,10,4.6,24,0,96
2026,January,1,USA,Retail,Home,Kitchen Blender,New,318,24860,9294,10,3625,26110,3.11,78.15,8,4.3,24,0,98
2026,January,1,Canada,Ecommerce,Grocery,Organic Snack Box,Loyalty,488,20540,7338,6,3310,55980,5.02,40.88,7,4.5,29,0,100
2026,January,1,Mexico,Retail,Fashion,Running Shoes,New,255,28910,10482,14,2980,21450,2.68,113.24,9,4.2,20,0,97
2026,February,2,USA,Ecommerce,Beauty,Skincare Kit,Loyalty,346,38250,14520,11,4375,44120,3.58,110.52,6,4.7,25,0,95
2026,February,2,Canada,Retail,Electronics,Smart Speaker,Returning,221,26380,9824,9,2815,18040,2.42,112.06,5,4.4,26,0,99
2026,February,2,Mexico,Ecommerce,Home,Air Purifier,Returning,279,30170,11188,13,3565,32740,3.08,108.18,7,4.3,21,0,96
2026,February,2,USA,Retail,Grocery,Protein Snacks,New,395,22140,7682,7,3090,29880,4.26,56.05,8,4.1,25,0,94
```The goal is to keep the same column schema as the original CSV so the agent can append new rows without touching previous records and then recalculate KPIs, charts, and filters without losing consistency.
You can also replace the entire dataset if you need to load a completely new CSV from scratch. In that case, use the same flow but make it explicit that you want to replace the whole previous dataset.
As additional examples, you can use these prompts when you want to replace the dataset or inspect snapshots, without treating them as part of the step-by-step flow:
Replace the `Retail Analytics Overview` dataset completely with this new CSV and update the dashboard.Return the dataset to the previous snapshotList the snapshots for the `Retail Analytics Overview` dashboardBy default, Luminon keeps up to `10` snapshots per dashboard. If you need more history, you can change that limit by setting the `LUMINON_DASHBOARD_SNAPSHOT_LIMIT` environment variable before starting Luminon.
LUMINON_DASHBOARD_SNAPSHOT_LIMIT=25 luminon start