Asking Vega and investigating energy performance
Refine answers with follow-up questions
Table of contents
Continue when the objective is still the same
Vega retains the messages and analytical context in a conversation. Use a follow-up when you are still investigating the same site, period, or decision. Start a new conversation when the subject or intended audience changes substantially.
A useful sequence moves from a broad result to validation and then explanation:
- Establish the total or trend.
- Break it down by time or component.
- Find the largest changes or unusual intervals.
- Verify coverage, units, and comparison boundaries.
- Ask for possible explanations and next checks.
Narrow the result
Refer clearly to the previous answer and state what should change:
- “Use the same component, but include only 1–31 August 2026.”
- “Keep the period and show daily values instead of one total.”
- “Show only the five highest hourly load values.”
- “Exclude the incomplete current day and recalculate.”
For important results, repeat the @ site or component reference rather than relying only on “it” or “that meter.”
Change one analytical dimension at a time
Make comparisons easier to verify by changing one element per follow-up:
- Keep the same scope and change the period.
- Keep those periods and change the granularity.
- Keep the same data and request a table or chart.
- Only then ask for interpretation or recommendations.
Example conversation:
Show daily consumption for @Main Warehouse / Grid Meter during August 2026.
Compare those daily values with July 2026 and show the percentage change for each complete day.
Rank the five largest increases and state whether any missing intervals affect the ranking.
Ask Vega to expose its basis
When a result is surprising, do not begin with “Why did this happen?” First establish what the data actually shows:
- “Which measurement definition and unit did you use?”
- “State the exact start and end timestamps.”
- “List the components included in the total.”
- “How many intervals were available and were there gaps?”
- “Show the values behind the conclusion in a table.”
After the basis is clear, ask Vega to separate evidence from hypotheses:
List plausible operational explanations for the increase. For each explanation, say what additional data or check would confirm it. Do not present a hypothesis as an observed fact.
Request a different presentation
You can ask Vega to turn the same prepared result into:
- a concise management summary,
- a detailed technical explanation,
- a table,
- a line or bar chart,
- a component breakdown, or
- a ranked list of highest or lowest periods.
Keep the metric and aggregation explicit. For power, say whether each bucket should show maximum, minimum, or average kW. For energy, state the time interval over which kWh should be summed.
Correct a misunderstanding
State the correction directly and ask for a fresh calculation:
The previous result used the whole site. Use only @Main Warehouse / Grid Meter, keep the same dates, and recalculate daily consumption.
Do not merely ask Vega to “try again”; identify what was wrong. Then apply the checks in Verify Vega’s answers before acting.
Stop an unhelpful response
While a response is being generated, select Stop response. Vega records the turn as Response stopped. Send a corrected or narrower request afterward. If the stop request is not acknowledged, use Retry stop when it becomes available.
A stopped response is incomplete and should not be used as evidence.