Own the number, and be able to defend it every month
The transformation office reports a number it did not produce, from teams that do not live in its tool. Lodestar computes status from dates and approved evidence, scores how much each report can be trusted, and lets workstream leads update the plan by replying to an email.
Green has to be earned
Status is derived from dates, approvals and evidence. A workstream cannot type itself green, and a green that outruns its evidence is flagged as a watermelon.
- Ten typed detectors: stagnation, burn mismatch, ghost projects, ignored risks
- Data-trust score per project, computed deterministically
- Evidence cadence enforced per KPI and per gate
No re-entry for the plan to move
A workstream lead sends the update they already write — PDF, Word, Excel, PowerPoint or plain text. The agent extracts typed edits and a human approves them, with permissions and audit identical to the manual route.
- Progress, dates, risks and resource gaps extracted from the document
- Every edit reviewed by a person before it lands
- Hash-chained audit entry per applied change
The monthly pack stops being a project
One calculation feeds every forum and every format, so the ExCo deck, the board pack and the live dashboard cannot disagree — and nobody spends a night assembling them.
- PDF, PPTX, DOCX and dashboards from the same live model
- Decisions taken in the forum link back to the plan item they move
- Overdue actions chased and escalated automatically
RUNS ON
EVIDENCE GATESDATA-TRUST SCORINGANOMALY ENGINEAI EXTRACTIONREPORT STUDIO
SAME ENGINES THE SCREENS RUN ON
The transformation office toolkit
RAG derived from dates, approvals and evidence — never hand-typed.
Data-trust scoring exposes projects green outside and red inside.
Agents extract typed plan edits from the update document, for human approval.
Finish dates projected from the performance trend, weeks before the miss.
Board, SteerCo and ExCo packs in four formats from one calculation.
Every applied change sealed on a hash chain with its approver.
See it on your own program, with your own numbers
A 45-minute working session: we load a slice of your portfolio and show the roll-up, the exceptions Polaris finds, and the pack it generates.