Skip to content
Business BenchA benchmark for AI agents on business work

Tasks·desk·Extraction

Abstract the plaza leases for the lender

lease-abstracts·time budget 1800 s·source

The request

The task request supplied to the adapter, alongside the workspace below. Runtime instructions, authorized tools and skills also affect the system. Containerized runs withhold host-side references; local mode is not an isolation boundary.

The bank wants lease abstracts for everyone in the plaza before the refinance closes. Go through the leases folder and put them in lease_abstracts.csv - Marisol's note says what she needs.

Workspace

  • leases/104_rent_schedule_exhibit.pdf
  • leases/Lease_102_executed.pdf
  • leases/Suite101_Lease_Basic_Provisions.pdf
  • leases/Suite108_lease_riders.pdf
  • leases/Suite110_lease_excerpt.pdf
  • leases/Suite_106_Lease_2019.pdf
  • leases/scan_suite_106_lease_summary.pdf
  • note_from_marisol.txt

Expected deliverables

  • lease_abstracts.csv

Recorded acceptance checks

The table uses check names and outcomes from the frozen campaign ledger. Acceptance requires all required predicates to pass; a predicate may itself contain a tolerance or coverage threshold. Its scope is part of the measurement contract.

Recorded checkTypeRequiredFailed / observed
requested columnscsv_columnsyes0 / 6
one row per suitecsv_set_equalyes0 / 6
tenant namescsv_values_matchyes0 / 6
lease datescsv_values_matchyes1 / 6
rent in effect on 1 October 2026csv_values_matchyes0 / 6
escalation typecsv_values_matchyes0 / 6
escalation amountcsv_values_matchyes0 / 6
option count and CAM capcsv_values_matchyes0 / 6
option notice deadlinecsv_values_matchyes1 / 6
expired lease left outtext_not_containsyes0 / 6
Current task declaration (10 checks)

From task.yaml. Original runner grading and frozen retrospective scoring are separate. Read the frozen source for the rules behind the reported results.

  • requested columns — csv_columns, required
  • one row per suite — csv_set_equal, required
  • tenant names — csv_values_match, required
  • lease dates — csv_values_match, required
  • rent in effect on 1 October 2026 — csv_values_match, required
  • escalation type — csv_values_match, required
  • escalation amount — csv_values_match, required
  • option count and CAM cap — csv_values_match, required
  • option notice deadline — csv_values_match, required
  • expired lease left out — text_not_contains, required

Recorded results, desk release

pass pass, frozen scorer only fail timed out
SystemRepetitionsFrozen / 3Original / 3Median minutesMean model estimate (USD)
Proto / DeepSeek V4.1 Flash222.60.0306
Codex / GPT-5.6 sol331.80.2394

These are repeated executions on the released fixture, not independent new business cases. Cost is captured model usage priced under the recorded assumptions, excluding infrastructure and human review. Per-attempt check verdicts, raw and frozen verdicts, usage, receipt and artifact hashes for this task are in results/latest/attempts.jsonl under "task": "lease-abstracts".

Recorded results, desk comparison, 28 September 2026

SystemRepetitionsFrozen / 3Original / 3Median minutes
Proto / GPT-6 sol331.8
Codex / GPT-6 sol331.4
Proto / DeepSeek V4.1 Flash331.0

A separate campaign under the same frozen scorer, not pooled with the release. Conditions and data.