I'd like to hand some spreadsheet cleanup work over to Buda, but there's one detail I'm thinking about: a blank cell can mean someone hasn't filled it in, or that the data isn't available. That's different from zero.
Do you explain all those rules before starting, or first ask the Agent to check which columns need clarification?
I'd prefer it to flag anything uncertain rather than fill in zeros just to complete the spreadsheet. Does anyone have a good way to phrase that?
Yes, but I would let the Buda Agent inspect a sample first, then agree on column meanings with the owner rather than trying to document every exception up front. The first invariant is simple: zero is a confirmed number; blank may mean missing, not entered or not applicable, and must not be silently changed to zero. For each important column, note when zero is valid, what blank might mean, and who can resolve it. Keep the source row and original value for any proposed change.
For example, zero sales can mean no purchases were made; blank sales may mean orders have not been reported yet. Ask the Agent to profile ten sample rows, list blanks, zeros and N/A values by column, and propose a small data dictionary. After you confirm the meanings, it can clean only the unambiguous rows and put the rest in a review list. Report confirmed zeros separately from unknowns so averages and completion rates are not distorted.
A prompt: “Read the headers and ten sample rows. Identify columns where blank, 0 or N/A is ambiguous. Preserve original values and row numbers; do not fill blanks with zero. Show me the proposed meanings and uncertain rows before editing the full sheet. For each changed value, record its source.” Save the dictionary as a Skill if this recurs. Which columns cause the most trouble: money, counts or dates? The Chinese discussion covers the same case.
Sign in to join the discussion. Reading is always open; you only need an account to post.
Sign inI'd like to hand some spreadsheet cleanup work over to Buda, but there's one detail I'm thinking about: a blank cell can mean someone hasn't filled it in, or that the data isn't available. That's different from zero.
Do you explain all those rules before starting, or first ask the Agent to check which columns need clarification?
I'd prefer it to flag anything uncertain rather than fill in zeros just to complete the spreadsheet. Does anyone have a good way to phrase that?
Yes, but I would let the Buda Agent inspect a sample first, then agree on column meanings with the owner rather than trying to document every exception up front. The first invariant is simple: zero is a confirmed number; blank may mean missing, not entered or not applicable, and must not be silently changed to zero. For each important column, note when zero is valid, what blank might mean, and who can resolve it. Keep the source row and original value for any proposed change.
For example, zero sales can mean no purchases were made; blank sales may mean orders have not been reported yet. Ask the Agent to profile ten sample rows, list blanks, zeros and N/A values by column, and propose a small data dictionary. After you confirm the meanings, it can clean only the unambiguous rows and put the rest in a review list. Report confirmed zeros separately from unknowns so averages and completion rates are not distorted.
A prompt: “Read the headers and ten sample rows. Identify columns where blank, 0 or N/A is ambiguous. Preserve original values and row numbers; do not fill blanks with zero. Show me the proposed meanings and uncertain rows before editing the full sheet. For each changed value, record its source.” Save the dictionary as a Skill if this recurs. Which columns cause the most trouble: money, counts or dates? The Chinese discussion covers the same case.
Sign in to join the discussion. Reading is always open; you only need an account to post.
Sign in