Energética

Examples

Six real use cases, each showing the natural language prompt and the tool the agent picks.

1. Track Vaca Muerta production trends

Prompt: "Show me Vaca Muerta oil and gas production by year since 2018."

Tool: query_production with formacion: "VACA MUERTA", fecha_desde: "2018-01", agrupar_por: "anio"

Why it matters: Vaca Muerta is the global shale play to watch outside the US. Year-over-year growth tracks Argentina's macro story.

2. Find top wells by company

Prompt: "Top 10 YPF wells in Neuquen by cumulative oil production."

Tool: query_wells with empresa: "YPF", provincia: "NEUQUEN", limit: 10

Use case: Investment due diligence, asset valuation, M&A research, benchmarking operator performance.

3. Compare WTI and Brent over time

Prompt: "What was the monthly average WTI price in 2024?"

Tool: query_prices with serie: "WTI", fecha_desde: "2024-01-01", fecha_hasta: "2024-12-31", frecuencia: "mensual"

Use case: Trading desks comparing intl benchmarks against Argentine local crudes (Escalante, Medanito).

4. Upstream investment by basin

Prompt: "How much have operators invested in Cuenca Neuquina drilling since 2020?"

Tool: query_investments with cuenca: "NEUQUINA", concepto: "perforacion", anio_desde: 2020

Use case: Energy analysts, banks doing project finance, journalists covering the energy beat.

5. Argentina energy trade balance

Prompt: "Argentina's crude oil exports in 2024."

Tool: query_trade with anio: 2024, flow: "export", producto: "crude"

Use case: Macro analysts modeling trade balance impact, FX impact, tax revenue.

6. Custom SQL analysis (Professional+)

Prompt: "For each basin, what's the ratio of horizontal to vertical wells drilled in 2024?"

Tool: execute_sql with a query like:

SELECT
  cuenca,
  COUNT(*) FILTER (WHERE tipo_pozo ILIKE '%horizontal%') as horizontales,
  COUNT(*) FILTER (WHERE tipo_pozo ILIKE '%vertical%') as verticales
FROM pozos
WHERE EXTRACT(YEAR FROM fecha_primera_prod) = 2024
GROUP BY cuenca
ORDER BY horizontales DESC

Why SQL: Curated tools cover the common cases. When you need cross-table joins, window functions, or custom aggregations, drop down to SQL.

Tips for getting good answers

  • Start with get_data_freshness if you're not sure how recent the data is.
  • Use get_schema when the agent needs to know which columns exist.
  • Be specific about dates — without filters, queries often hit the row limit on tables with millions of records.
  • Argentine basin names are uppercase in the data (NEUQUINA, GOLFO SAN JORGE).
  • Currency: production is in m³ and miles m³, prices are in USD/bbl, investments are in MMUSD, trade is in FOB USD.