Beta Not the live site. Pages in review Live site →
Loft

The Tokenator  /  AI spend, costed

AI made a new line item. LOFT made it a number.

The production platform for AI animation and VFX: bid, crew, schedule, generate, review, cost. The Tokenator tracks model spend against the bid, not the invoice.

Tokens · The Keeper's Light Spending ahead of progress
Budget spent87%
Work approved70%
124% of budget at completion
≈ 26,700 of 21,500 tokens
SpentEarned by approved workForecast
Allowance Zoe Park 5,880 / 6,000 Hard
SH030 · take refused Needs 210, 120 left Nothing spent. Ask a manager to raise it.

You know what the crew costs. Now you know what the tokens cost.

Budget & Track AI Token Spending, the tutorial

The tutorial · 6:50

Don't take our word for it

Everything on this page is in one recorded session of the real product. An AI short, The Keeper's Light, is budgeted from its bid, run for three weeks, and costed on its P/L. Each step below links to its moment in the video.

The Keeper's Light is demo data, made for the tutorial.

How it works

Bid it before you generate it

The bid, with a Tokens column next to the days per department
01 · The bid

Tokens are a column in the bid

Next to the days each shot needs in each department, the bid holds the tokens it's expected to spend. A ten-second video take costs 420; a frame, 10 to 25. The show's total is right there with the labour.

See it at 0:48 →
Costing, by asset: a budget per shot, character and location, rolled up by sequence
02 · Publish

The bid becomes the budget

Publish the bid and every shot gets its own token budget, rolled up by sequence into one for the project. Each budget is soft, flagged when it's passed, or hard, refused.

See it at 1:19 →
Costing, by user: Zoe Park with a hard allowance of 6,000 tokens
03 · People

An allowance for each artist

Before anyone starts, give the people who generate an allowance. Make it hard and a take that would go past it doesn't run until a manager raises it.

See it at 2:30 →
Progress: tokens spending ahead of progress, a forecast of 124% of budget and the token burn chart
04 · Progress

Know in week three, not week six

Spend is measured against the work approved, not just against the budget. Three weeks in, 87% of the tokens are spent but only 70% of the work is approved, and the forecast says the show finishes at 124% of its budget.

See it at 2:51 →
A take refused before it runs: the allowance can't cover it, with who can raise it
05 · The limit

Refused before it runs

The take needs 210 tokens and the artist has 120 left, so it's stopped with the reason and who can change it. Nothing is spent. The producer asks the Assistant for another 1,500 tokens and the work carries on.

See it at 3:55 →
The Project P/L with AI generation as its own line next to labour
06 · The P/L

A line on the P/L

AI spend sits on the project's P/L as its own line, next to labour. Token spend is costed like every other line, instead of discovered like a bill.

See it at 4:37 →

Three levels

Every token, against every bid

The project

The bid's total. Spend, what the approved work earned, and a forecast at completion.

Every shot and asset

Its own budget from the bid, with what's spent and what's left, rolled up by sequence.

Every person

An allowance for whoever generates. Soft to flag it, hard to stop it.

Everyone else sold you the credits. LOFT tracks what they cost the show.

Meet the Tokenator

The first AI spend tracker built into a production platform.