Traditional sprint tools count one thing: human hours. We count two. Every story is routed at planning time — AI-assisted or human-owned — giving you accurate burndowns, honest capacity tracking, and a complete audit trail the moment planning closes.
At the start of each sprint, every story is classified through five signals and routed — AI-assisted, human-owned, or either. The board reflects actual execution paths, not one-size assumptions.
Every story is evaluated against five signals at the moment it enters the sprint. The classifier scores each signal and outputs a routing recommendation — not a mandate. Human override is always one click away.
Are the acceptance criteria objectively checkable — pass/fail, schema diff, test output? Objective = AI-eligible. Requires subjective judgement = human-owned.
Auth, payments, PII, production infra — human-owned regardless of how mechanical the change looks. Risk classification overrides all other signals.
Isolated to one module with well-tested boundaries = AI-assisted. Crosses multiple modules or requires coordinated refactoring = human-owned.
Resembles successfully completed prior work = AI-assisted. Net-new for the team, unfamiliar domain, or requires creative problem-solving = human-owned.
Anything end-users see directly — copy, UI flows, error messages — gets a human owner even if the implementation is technically AI-eligible.
Traditional sprint tools track one number. A team running AI-assisted delivery has two non-interchangeable capacity pools — and treating them as the same number produces forecasts that are wrong before the sprint starts.
We track both dimensions independently: human hours (with approval overhead reserved) and AI-assisted run budget (with token cost visible). Burndowns, velocity, and cycle-time stats are all computed per dimension. You always know exactly what is left and what is drawing on it.
Every routing decision, every approval, every mid-sprint reroute is a logged, immutable event with full attribution, timestamp, and context. The audit trail is live and inspectable at any point during or after the sprint.
Enterprise compliance teams get exactly the artefact they have been asking for since agile became the norm — a complete, attributable, exportable record of what was done, by whom, when, and why.
Plans that route each story correctly on day one produce burndowns that track reality instead of diverging from it. +38% forecast accuracy measured across 12 consecutive sprints.
Routine tasks that previously consumed full-day engineer slots complete in hours. The freed capacity flows to complex human-owned work — giving 1.7× throughput without adding headcount.
Every mid-sprint reroute, every routing outcome, every velocity measurement feeds back into the classifier. Routing accuracy increases sprint-over-sprint — the system learns from actual results.
RBAC, full audit coverage, exportable records, and compliance-ready approval gate logs — no additional configuration required for your governance or security review.