Schedule risk and critical path in Jira (Advanced Roadmaps)

Jira earned its position. It is the dominant tool for agile software development, and Advanced Roadmaps is a serious planning layer: it rolls up initiatives over epics over stories, models cross-team dependencies, runs an auto-scheduler that respects team capacity and sprint cadence, and lets you compare scenarios before committing a plan. For "teams of teams" coordinating delivery at scale, that breadth is real and hard to replicate.

The ecosystem compounds the advantage. The Atlassian Marketplace, deep integration with Bitbucket, Confluence, and Jira Service Management, and the gravity of an org already standardized on Jira mean the switching cost is high and the surrounding tooling is mature. CritPath does not compete with Jira on agile delivery or marketplace depth — it competes on the schedule-risk math Jira was never designed to carry.

The gaps are structural. Advanced Roadmaps has no native critical path analysis — you can sequence and auto-schedule work, but Jira will not tell you which chain of dependent work actually drives the end date, or how much float each path has. There is no Monte Carlo simulation, so there are no probabilistic finish dates: every roadmap bar is a single deterministic guess, exactly the kind of false-precision date that gets carried into a board deck and then missed.

Rovo, Atlassian's AI, is capable — but it reasons over the work graph (issues, statuses, conversations), not over a schedule network. It cannot tell you which task is driving your P80 slip, what a 50%-confidence buffer should be, or what happens downstream when a gate decision lands, because none of those constructs exist in Jira. And the schedule-risk methods that R&D programs lean on — Critical Chain, Theory of Constraints / Drum-Buffer-Rope, decision gates, WSJF and Cost of Delay, AACE-grade risk-driven scheduling — are simply absent.

CritPath AI is purpose-built for the job Jira leaves open. It runs a full Critical Path Method (four dependency types, lag, float, near-critical analysis), a Monte Carlo engine (PERT-Beta distributions, risk-event injection, criticality index, tornado sensitivity, P50/P80/P90, optional duration correlation), CCPM with Theory of Constraints and Drum-Buffer-Rope (drum, project and feeding buffers, fever chart), and decision gates with retroactive rescheduling so a Go/No-Go/Pivot/Defer at a TRL or CMC milestone re-cascades the whole plan.

It also carries the methods R&D programs prioritize by: WSJF and Cost of Delay, EVM (SPI/CPI/EAC), and AACE RP 132R-23 Level 4 risk-driven scheduling with an append-only audit log. The differentiator is the copilot: built on Claude and Gemini and grounded in your actual dependency graph, it can tell you which task is driving your P80 finish and what a gate decision does downstream — schedule-aware reasoning, not a chatbot pinned over a backlog. It runs in the browser with multi-tenant orgs, 2FA, and XLSX/CSV import — including from a Jira export.

For most R&D teams this is not a rip-and-replace decision. Jira stays the system of record for engineering execution — sprints, issues, code, and the day-to-day flow of delivery work. CritPath AI sits above it as the program-level schedule-risk layer: import the milestone structure, model the dependency network, run the Monte Carlo, set the gates and buffers, and report the probabilistic date the board actually needs.

That split mirrors how the program runs in practice. Engineers live in Jira; the program lead, the VP of Development, or the PI lives in the risk model. CritPath gives the second group the CPM, CCPM, and probabilistic dates Jira does not — without asking the engineering org to leave the tool it already runs on.

The table below compares the two on the capabilities that decide a schedule-risk purchase for an R&D program. Jira pricing is per-user list pricing and varies by edition and seat count; CritPath is a single flat seat with every standard feature.

If your need is agile software delivery — coordinating many engineering teams, managing a shared backlog, and integrating tightly with your dev toolchain — Jira with Advanced Roadmaps is the right tool, and CritPath is not trying to replace it there. Keep Jira for execution.

But if you need a defensible probabilistic finish date, a real critical path, CCPM buffers, decision gates that re-cascade the schedule, and an AI copilot that reasons over the actual dependency graph, Jira simply does not have those constructs. CritPath AI occupies that quadrant — the schedule-risk engine modern generic PM tools never built — at $10/user/month, with AI usage billed separately and metered. The two are complements: Jira runs the work, CritPath quantifies the risk in the date.