The Client's Claude Does the Work

In v1 we pay for almost no model calls. The partner's own Claude writes the memo; MemoPop supplies the method and keeps the work.

Path
decisions/The-Clients-Claude-Does-the-Work.md
Authors
Michael Staton
Augmented with
Claude Code on Claude Opus 5.5 (1M context)
Tags
Decision · MemoPop · Claude-Connector · Pricing-And-Cost

The Client’s Claude Does the Work

The decision

In v1, the client’s own Claude (Desktop, web, or mobile) does all the reading, research, and writing. MemoPop’s server supplies the method, holds the deal’s materials and artifacts, versions them, backs them up, and compiles the finished memo.

v2 adds the server-side pipeline: our own model calls, paid through provisioned tokens or the client’s own keys, and our research APIs (Perplexity, Tavily, and the rest).

Decided 2026-10-10 by Michael Staton.

Why

  • The value is the method, not the model calls. Months of iteration produced a staged method that writes better memos than any the operator has seen, and better than anyone they’ve sent them to has seen. Packaging that method is the product; v1 is a staged instruction set for better output.
  • Clients already pay for Claude. Running on their subscription means we meter almost nothing, so billing can wait.
  • Partners are already in Claude all day. No new app to install or learn.

What this rules out for v1

  • A “generate the whole memo” job on our server.
  • Research APIs on our keys. In v1, research is Claude’s own web search plus what the partner adds.
  • Payments and token provisioning, which stay with the later didi.sh billing work.
  • [[MemoPop-as-a-Claude-Connector]] — the exploration this came from
  • [[The-Server-Hands-Claude-the-Method-One-Step-at-a-Time]] — how the method reaches Claude