Pitch 04 · Kevin's proposal

Inference Partner

Make Something Big the inference partner for the DATA platform. DATA is the harness and the skills; Something Big's assistants and the models behind them do the thinking and the errands. Every DATA seat becomes a Something Big seat.

Steve: "clearly you can run DATA"
How one job flows

Tap through the drawing.

PERSON DATA SKILL INFERENCE ASSISTANT RECEIPT SHEET 04 · DRAWN BY KEVIN · NOT TO SCALE · PROPOSAL

1 · Ask

A person on a DATA team texts or emails their Something Big assistant: "Pull last month's churned customers and draft win-back emails."

2 · Pick a tool

The assistant reaches for a DATA interface (a skill) for that job. DATA installs on any platform that takes Anthropic-style skills or MCP, per Steve.

3 · Think

The model work runs on Something Big's side: the assistant's own computer and the inference behind it. This is the part Something Big would be paid for.

4 · Do

The assistant does the work with its own email, phone and computer, and asks before anything that spends money.

5 · Pay the maker

DATA pays the skill's creator 50% of the compute each time it's used (Steve, on Tokenized). The person gets the result and a receipt.

The flow is Kevin's sketch of how the two could fit, not a description of a live integration.

Credits, in public numbers

How much work a partner team buys.

Something Big revenue per year
Credits per seat, per month
Seats
Credits per month, all teams
Revenue per month

What a credit buys (which models, how many tasks) isn't stated here. Sliders, not a forecast.

Proposal, not agreed

The first step

  • Steve: pick one DATA interface to run through a Something Big assistant end to end.
  • Matt / Nate: say what DATA would need to call Something Big inference (an account, an API, a plan).
  • Kevin: run that one job and publish the log and the cost.