‹ Product recipesProduct · Agent Blueprint
Onboarding a new PM or engineer
What did we decide about each subsystem? Every decision and rationale, chronologically, cited.
Side by side: token usage, with and without Paperbase
~91k tokens saved · 76% less work for the agentWithout Paperbase
~120k tokens
- Search docs for the subsystem and read everything that matches.
- Ask the previous owner what was decided and why.
- Skim old PRDs and review threads to infer rationale.
- Re-read meeting transcripts from months ago.
- Reproduce the decision history from whoever happens to remember it.
With Paperbase
~29k tokens
- Pull every Decision attached to {{subsystem}}, in order.
- Return the Rationale alongside each.
- Pull the interactions that led to the key decisions.
- Compose the subsystem history as one narrative.
Agent prompt
You are bringing a new PM or engineer up to speed on {{subsystem}}.
Using Paperbase:
1. Pull every Decision made about {{subsystem}}, in chronological order.
2. Return the Rationale for each, and who made it.
3. For the 3-5 pivotal decisions, pull the interactions that led up to them.
4. Note any currently open questions or reversals.
Output:
- A chronological decision history of the subsystem.
- Each decision: what, why, who, when, and the citation to follow.
- Open questions and unresolved tensions.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.