待翻译：Project management with schedule aware AI copilot

AI 服务暂时不可用，以下为来源正文，待恢复后补全翻译。

CritPath AI — Schedule Risk Management & CCPM Software for R&D LearnCompareSolutionsPricingArchitectureFAQAboutSign in SCHEDULE-RISK PLATFORM FOR R&D PROGRAMS Give your board a date you can defend. CritPath AI runs real schedule math over your actual plan — so you know how solid every date is, see trouble weeks early, and the plan reschedules itself when a vendor slips. Built for biotech, deep-tech, and federally funded R&D. Join the waitlistSee how it re-plans a slip $10/user/month — every feature. AI employees metered at cost + margin. No credits. Built for Biotech program leaders · Deep-tech founders / CTOs · Federally-funded PIs · Grant program managers A vendor slips. Your plan re-plans itself. The re-plan that used to cost your PMO a day happens before the meeting ends. 65 seconds, no cuts: a tox-study risk is marked Occurred → the schedule re-plans retroactively → the buffer fever chart and the P90 date move. A CMC vendor slips a month. A risk you logged just hit its trigger. Mark it occurred — that's all the input CritPath needs. Every downstream task, buffer, and milestone re-cascades in seconds. The engine re-runs the full schedule math automatically — nothing is hand-dragged. New dates on the Gantt, buffers re-checked — and on projects running Monte Carlo, a fresh read on how solid the new finish is. Ready for the meeting you're already late for. Sound familiar? Dates nobody can defend Your Gantt shows one date and says nothing about how likely it is — so every board deck is a promise you can't back. CritPath shows how solid each date is and which tasks are quietly driving the risk. One slip, a week of re-planning A CMC vendor slips a month and every downstream milestone silently moves. Re-planning by hand costs your PMO a day each time. Here, you log the slip and the plan catches up in seconds. Reporting is a second job One program PM spent eight hours a day re-typing emailed status updates. Your status report already exists — it's the live schedule. Export it, or have an AI employee draft it, every number computed. The engine R&D actually needs Built from the first line for the uncertainty and decision density of research programs — not retrofitted from construction or IT delivery. Know how solid every date is Monte Carlo simulation — 10,000 runs over your actual dependency network — turns a single guessed date into a defensible range and the real chance of hitting your target. Your engineers already estimate in ranges; here the range survives all the way to the board deck. Find what's eating your finish date Critical path plus critical chain (CCPM): the bottleneck, the buffers protecting your milestones, and a fever chart that shows margin burning before the date moves. Decisions that live in the plan Go / No-Go / Pivot gates are schedule objects with named owners, co-signs, and an audit trail — and a No-Go or Pivot triggers a fresh run of the schedule math. Staff it to finish sooner Resource leveling plus a staffing optimizer that shows what each hire or reassignment buys you — the speed-versus-cost trade-off, priced in days and dollars. AI employees, governed Configurable agents read the live schedule and propose updates for your review. Nothing touches the plan until a named human accepts — every acceptance audited. Regulatory-grade rigor AACE RP 132R-23 Level 4 risk-driven scheduling with a tamper-evident audit trail — built for pharma IND programs and federally funded R&D. 21 CFR Part 11 on the Enterprise roadmap. AI employees. Proposes. You approve. Build AI employees — an Estimator, a Risk Analyst, a sub-PM — with skills you define. They read your live schedule, run grounded analysis, and return typed proposals — you review the proposed change, not pages of prose, and accepting takes seconds. Nothing lands until a named human accepts; every acceptance is audited, and when you enable the adversarial reviewer on a persona, a separate, isolated review pass — blind to the generation context — grades every run against your own quality rubrics, and blocking failures prevent acceptance. Autonomous AI-employee runs activate at general availability. Billed per completed job at its estimated token cost plus a stated margin — no credits, no black-box meter. An AI employee holds no seat and no subscription of its own: you pay for the work it actually finishes, and nothing when it is idle. See how AI usage is billed. ✓Decision gates as first-class schedule objects — owned, co-signed, audited — no PMO tool ships this ✓CPM + CCPM/TOC + Monte Carlo in one engine — incumbents have at most one ✓Every AI-employee change lands as a human-approved, audited proposal against the plan ✓AI usage metered per completed job at estimated cost + a stated margin — no credits, no seats, no black-box meter ✓WSJF + Cost of Delay wired into the simulation and the staffing optimizer ✓TOC Thinking Tools — Goal Trees, Conflict Clouds — generated from the live schedule The gap no one else fills Risk-math depth →Modern, self-serve UX → The empty quadrant Legacy risk tools Primavera · Deltek · Pertmaster — ~$10K+/seat, Windows desktop Spreadsheets & trackers manual Gantts — no probabilistic math Generic PM Monday · Asana · Smartsheet · Linear — $7–25/mo CritPath AI Real risk math + modern UX + a schedule-aware copilot $10/user/mo · all features Schedule-risk tools have the math but not the UX. Generic PM tools have the UX but not the math. CritPath AI is the only one in both. How we compare Rigor you can check No borrowed logos, no invented percentages — claims you can verify in the product and the literature. Vectorized Monte Carlo over your full dependency network — with sensitivity ranking of which tasks actually drive the finish date. The resource-leveling engine is validated against published academic scheduling benchmarks (RCPSP/PSPLIB) — not a heuristic we eyeballed. The cost of delay in late-stage pharma (Tufts CSDD). CritPath prices delay into prioritization and staffing decisions via Cost of Delay + WSJF. Per-organization append-only hash chain with in-app integrity verification — the AACE Level 4 foundation, shipped today. Pricing One simple seat — $10/user/month, every feature. The market’s own alternatives are a $149 desktop add-in or a $10,000-a-seat analyst tool; this is the engine in between. Pro Most teams $10 /user/mo Everything: CPM, TOC/DBR, Monte Carlo, decision gates, WSJF, portfolio dashboard, 2FA/MFA, audit log, API. Unlimited projects. Join the waitlist Enterprise Custom Dedicated CSM, training, custom integrations; on-prem, SOC 2, SAML, HIPAA BAA, and 21 CFR Part 11 on the Enterprise roadmap. Talk to us Build your own organization of AI employees — configurable agents with custom skill sets that work alongside your team. Billed separately by usage, metered per completed job: the estimated token cost of that job plus a stated platform margin. No credits to buy, no hourly minimum, no surprise burn — and a run that fails bills nothing. See full pricing, seat rules, and how AI usage is metered → Model your next milestone. Spin up a project, import your task list from CSV or Excel, and see how solid your dates really are — in under five minutes. Join the waitlistI already have one