Compare the commercial impact of a scope change before making a client commitment.
A local workstation application for reviewing fixed-fee agency project changes: source-backed scope comparison, deterministic margin and capacity scenarios, separate internal reviewer decisions, and controlled exports. Optional AI helps prepare drafts.
For small agencies running fixed-fee projects with a written scope and dependable cost records. Bring together the signed agreement, actual costs, remaining-work estimate, change request, capacity and margin policy. The five-step local workflow moves from evidence through baseline, scope comparison and scenarios to a separate internal reviewer's decision and export. It works without AI.
A Python engine compares baseline, absorb, quote, re-scope, defer and decline options using project fee, forecast direct cost, contribution margin, stressed margin and aggregate current-window capacity gap. It flags missing or stale citations and inconsistent ledger rows rather than inventing figures. It does not calculate full agency profit or detailed staffing schedules.
A separate named account records an internal review, but two accounts do not prove two people acted. Staff must check source truth, contractual meaning, estimates, client-safe wording and actual client acceptance. No quote, invoice, signature, email, staffing action or other external commitment is automated.
Evidence and limits: The package reports 50 automated Python tests and eight DOM workflow checks, plus limited fictional agent exercises. No measured customer outcomes, full real-browser or accessibility acceptance, Windows/macOS execution or native Claude qualification. It is single-workstation software, not a multi-company portal: all local accounts can see all projects, backups are not encrypted by the application, and optional AI sends files to a separately chosen provider under that provider's terms.
Before you buy: This is a working local product for a controlled pilot, not a validated broad rollout or a hosted enterprise service. Real customer benefits and target-device acceptance have not been established.
Requires Python 3.10+ and a browser on the same trusted computer. AI is optional and no AI subscription, hosting, setup service or external system integration is included.