Research
Intersecting planes of coloured light standing like a lit grid in the dark.

Note 04From the deskaccounting · 2 min read

The company dashboard

The quirq ledger a company reads monthly, and the reading discipline that goes with it. A worked quarter where token spend rose 83% while verified value per all-in dollar rose 81%.

Read it interactively
The calculus exists to produce one artifact: a ledger a company reads monthly the way it reads its P&L.

A worked quarter

Three unit types (support tickets, contract reviews, weekly reports) in a mid-size support-and-engineering operation. Every number computes from how to calculate it; parameters come from the micro-benchmark ranges of this research note. The table is arithmetic, not measurement: its role is to exhibit the full calculation chain, and the validation program's job is to replace it with production ledgers (experiment E5).

MonthUnitsPotential ($B)Minted QInferenceCompute+APIInterventionC_totalQER
April2,10018,40015,770$1,490$410$3,120$5,0203.1x
May3,40029,90026,310$2,210$630$3,640$6,4804.1x
June4,80041,30038,000$2,730$820$3,280$6,8305.6x

Trend read: QER 3.1 → 5.6 (+81%). Intervention rate 18.1% → 11.4%. Cost per quirq 0.32 → 0.18. quirq velocity +141%.

The same quarter in tokens alone: spend rose from $1,490 to $2,730, +83%, a number indistinguishable from waste. That contrast is the whole argument: the token bill says costs nearly doubled; the quirq ledger says verified value per all-in dollar rose 81% while human rescue fell a third. One is an expense line, the other is an investment case.

The reading discipline

  • QER answers: is the program paying.
  • QER* trend (audit-corrected) answers: is it improving honestly.
  • Intervention rate answers: can it be trusted with more, and its per-check decomposition names the next environment investment. In the worked quarter one support-ticket check (KB linking) drives most interventions; hardening that single check moves the whole company trajectory.
  • Cost per quirq, by unit type, prices each category against its human baseline: a contract review at c_q = 0.13 against a paralegal-hour baseline is a decision, not a vibe.

The buyer's read of the ledger. The seller's read is next: pricing agents as a builder.

Adapted from the XO research program · canonical version at docs.xo.builders

Keep reading