Budget narrative
Where spend went and why, reconstructed from every approval conversation. The reasoning behind the numbers, ready for board or audit.
Why this is memory-only. A budget shows numbers. It never shows the reasoning behind them - why this line was approved, who pushed for it, what tradeoff was accepted. That reasoning lives in approval conversations, and once the quarter ends it exists nowhere but people's memory. Only a memory layer that captured those conversations can reconstruct the narrative behind the spend.
Side by side: token usage, with and without Paperbase
Memory onlyWithout Paperbase
Not recoverable from spend artifacts alone
- Export the spend report by category.
- Ask each budget owner to explain their numbers.
- Skim approval emails if they survive.
- Reconstruct the reasoning from whoever remembers.
- Present numbers to the board with no story their auditor can verify.
With Paperbase
Persistent memory layer
- Pull the quarter's spend by category and owner.
- Pull the Decisions behind major lines, with rationale.
- Attribute approvals to their conversations.
- Compose the budget narrative with citations.
Agent prompt
You are building the budget narrative for {{period}}.
Using Paperbase:
1. Pull spend for the period, grouped by category and owner.
2. For each major line, pull the Decision and Rationale behind it.
3. Trace approvals to their conversations, quoted and dated.
4. Flag spend that had no recorded approval.
Output:
- The budget, told as a narrative: why each line exists and who decided.
- Approvals cited to their source conversations.
- The lines with no rationale on record, flagged for follow-up.Placeholders in {double_braces} are inputs the agent will ask for at runtime. Give the prompt to any agent connected to Paperbase, and the rest grounds in your own memory graph.