‹ Sales recipesSales · Agent Blueprint
Negotiation prep
What has the CFO said about pricing? Every time a specific contact has weighed in on price: quoted, dated, and sourced.
Side by side: token usage, with and without Paperbase
~65k tokens saved · 72% less work for the agentWithout Paperbase
~90k tokens
- Search email for the contact's name and "price", "cost", "budget". Skim years of thread excerpts.
- Search Slack for the contact's name. Pull messages that mention pricing, competitors, or the deal.
- Search meeting transcripts for sessions the contact attended. Re-read each for stance cues.
- Search the CRM for deal history with that account and infer the pricing thread.
- Reconstruct, from memory and fragments, every position the contact has taken.
- Guess at whether positions changed over time and when.
With Paperbase
~25k tokens
- Pull every Interaction the contact has been part of, across email, Slack, and meetings.
- Ask Paperbase for extracted stances: pricing, budget sensitivity, competitive pressure, timeline.
- Pull Decisions about pricing attached to this account, with Rationale and date.
- Return each stance quoted, dated, and sourced to the original interaction.
Agent prompt
You are preparing for a negotiation with {{contact_name}} at {{account_name}}.
Using Paperbase:
1. Pull every Interaction {{contact_name}} has been part of at {{account_name}} - email, Slack, meetings.
2. Return every extracted stance from this contact on: pricing, budget, competitors, and timeline. Quote the source interaction and include the date.
3. Pull any Decisions about pricing or concessions at this account, with the Rationale attached and who made them.
4. Flag where the contact's position changed over time, with the moment it changed.
Output a negotiation brief:
- Every pricing-related statement the contact has made, quoted and dated.
- Our side's decision history on this account, with rationale.
- Three likely pressure points, each grounded in a citation.
- One line: what to concede last.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.