Acceptance delay
—
Relative to the original due date; negative means early.
Toekyn → Qaieo / Synthetic replay
Evidence in.
A clearer view ahead.
Follow one synthetic manufacturing milestone from recorded work to delay and remaining-cost forecasts.
Toekyn generates the work-evidence data; Qaieo combines it with simulation data for ML, RL and evaluations. Yeki’s planned integration adds government API context to underwriting and financing workflows.
Saved outputs from Qaieo’s actual synthetic evaluation run. No live ingestion, camera inference or model execution occurs in this browser.
Assembly · final test
SYNTHETIC · SHADOW ONLYHistorical fixture dates, not a current project. All amounts USD.
Loading the synthetic replay…
01 / Toekyn evidence
Evidence at the cutoff
Verified quantity uses planned native quantity. Rework and dispute fractions use distinct planned roots; these measures must not be summed.
Inspect source lineage
Hashes identify saved research artifacts. They do not prove physical truth; this page does not verify signatures or real permissions.
02 / Qaieo evaluation
Verified Work™ evidence
Remaining incurred cost
—
Future in-scope incurred cost, not a financing amount.
Target-specific research intervals from synthetic calibration groups. No empirical calibration or joint coverage guarantee.
Evaluation on hold
No replacement forecast or favorable default is supplied.
03 / Financing boundary
Shadow mode. No action.Proposed terms unavailable · Execution disabled · Reward absent · Policy promotion blocked
Compare with the later outcome
An independent authored outcome ledger supplies the label later. It was not available to the selected prediction.
What this demonstrates
A traceable path.
Not a credit decision.
The replay uses the first test case in a fixed-seed synthetic run, selected by its identifier rather than its result. The run used 120 milestones from 30 synthetic companies with separate training, calibration and test partitions.
Changing the information set selects a saved fitted-baseline output. It is not a causal estimate of what adding a camera or changing funding would do. The evidence-hold controls are illustrations of the research boundary, not live permission enforcement.
RL requires executed actions, policy information and attributable rewards. This run provides frozen states only; it is not eligible as an RL transition. External training remains off.
Reproducibility and limitations
Inspect the public synthetic snapshot ↗The original source code revisions and report/model hashes are included. Only a sanitized synthetic case is published; there is no upload or partner-data connection.