The working files, already built.

Four registers in one, with ageing and overdue flags.

Inherent and residual scoring, live 5x5 heat map, priced exposure.

Power-interest quadrants, attitude gap, live scatter grid.

Validates one Accountable per row and flags every breach.

Level, parent and budget roll-up derived from the code alone.

A real Gantt in Excel - bars, progress shading, current week.

Committed, incurred, accrued and forecast in one report.

Monthly demand by person, over-allocation flags, histogram.

Only approved change counted; exposure reported separately.

Actions with live ageing, plus a proper decision log.

Next-due dates roll forward from your chosen frequency.

Lessons aimed at a named process and tracked to adoption.

A one-page status report a sponsor will actually read.

The four questions a charter must answer, and nothing else.

One change, priced in money and days, with a recommendation.

Minutes built around decisions and owned actions.

Closure that compares against the original baseline, not the last.

Full EVM: CPI, SPI, TCPI, four EAC methods, S-curve, verdict.

Baseline against actual, plus the cash position nobody tracks.

Rules of credit - the mechanism that makes earned value credible.

Earned against actual hours, with the productivity factor.

Commitment, VOWD, accrual and forecast - a real cost report.

Control accounts by resource type, owner and measurement method.

Three-point priced trends caught before they become invoices.

Drawdown against risk exposure, with a cover ratio.

The version history that makes a variance report defensible.

All 14 DCMA checks scored automatically against thresholds.

Monte Carlo inputs in the shape every risk tool expects.

Delay events, notice periods, claimed against determined.

Two owners per interface, with agreed-in-writing flagged.

What the estimate includes, excludes, and how accurate it is.

The programme equivalent of a basis of estimate.

Cost, schedule, risk and change in one monthly pack.

Twelve weeks with an artefact required every week.

Versioned prompts with their known failure modes recorded.

Weighted model selection with a built-in sensitivity test.

ROI, payback and a priority score discounted for data readiness.

Three runs per case, separating unreliable from simply wrong.

Cost per request, per year, and at ten times the volume.

Hypothesis-first experiment log that keeps the failures.

Eighteen controls mapped to the NIST AI Risk Management Framework.

One page of thinking before anybody writes code.

Datasheets-for-datasets, as AI governance now expects.

NPV, IRR, payback and benefits tracked after approval.

Committed and aspirational key results graded differently.

Outstanding commitment and accrual, straight into your cost report.

One page. A decision, the reasoning, and the ask.

The written argument that goes with the financial appraisal.