Model the work as it really is
Hierarchical WBS, milestones, deadlines, manual and auto tasks — or import your MS Project XML, or the spreadsheet you actually plan in, and keep everything. Or just describe the program and let the agent draft it.
Planning agent & schedule-risk platform for program managers
Describe the program. See a thousand futures. Commit to one.
Foreweave's agent learns your program before it writes a task, and grows a real schedule from it — tasks, links, calendars and all. Then it simulates every way that schedule could play out, so you stop defending a single date and start choosing among futures.
From seed to schedule
Plant it. Weave it. Let it grow.
Foreweave's deterministic engine does the cultivation: every committed edit recomputes the full critical path in under half a second.
Link with all four dependency types
FS, SS, FF, SF with signed lead and lag, six constraint types, real working-day calendars. Cycles are rejected; conflicts are reported, never silently "fixed".
Watch the critical path live
Slack, float, and the critical path recompute on every edit. Drag a bar on the Gantt and the whole downstream reflows the instant you release it.
The gardener
It learns the program before it writes a task
Most assistants take a paragraph and hand back a plausible list. Foreweave's agent profiles the project, interviews you for what's missing, and gets you to accept a planning guideline — how it intends to plan — before a single task exists. It remembers those decisions next session. Then it drafts through the same validated engine you use, into a scenario you review as a diff. Your baseline is never touched without your hands on it.
Off by default, per tenant and per account
Zero-retention endpoint
Names redacted in transit
Drafts only — never your baseline
"Three build sprints, certification after integration tests, launch can't slip past Oct 24. How confident are we?"
Your accepted guideline puts a gate after integration, so I drafted 14 tasks into a new scenario. With your three-point estimates: P50 Oct 21, P80 Nov 2 — certification is critical in 84% of runs. I have two ideas for pulling P80 inside Oct 24. Want to see them as scenarios?
→ 14 × create_task · run_simulation(5,000 iters, seed 4217) · every action auditedThe whole garden
Everything between the first task and the last milestone
Eight things it does. All of them in detail →
Four views you never get lost in
A Gantt with frozen task panel and sticky date header — the "freeze panes" you always wanted in a schedule — plus a keyboard-fast grid, a network diagram, and an executive timeline. Drag to create, move, resize either edge, all with live previews and undo.
Scenarios as overlays
What-ifs never touch your baseline. Compare finish dates and critical-path changes side by side, then promote the winner.
Monte Carlo as a first-class citizen
Three-point estimates or named distributions, thousands of iterations, S-curves, criticality index, tornado sensitivity — every run reproducible from its stored seed.
Teams, resources, and honest workloads
Assign people, see over-allocation flagged the moment it happens, level resources when you choose to. Earned value — planned, earned, and the indices — comes out of the same numbers.
Track what actually happened
Baselines that stay put and render as ghost bars, actuals that anchor the engine above every other rule, a status date and a sweep — and a review that tells you what to chase before the update meeting starts.
Programs, not just projects
Roll member schedules into a program master under read-only mirrors. Changes publish upward; anything that moves the master's critical path waits in a review queue with an impact preview instead of landing behind your back.
MS Project and the spreadsheet
Round-trip XML in and out — plus the CSV and XLSX everyone actually plans in, including schedules drawn as coloured cells, read back with cell-level provenance you confirm before it becomes a plan.
Isolation your IT will believe
Your organization is a hard boundary, enforced twice — authorization before role, and row-level security in the database that fails closed. Everything in the interface is on the REST API too.
Two branches, one decision
Compare futures the way you'd compare anything else: side by side
A real comparison from a plan like yours — same program, one decision apart.
Baseline · single test lab
P80 misses launchNov 4
- P50 finishOct 21
- P80 finishNov 4
- Certification criticality84% of runs
- Critical pathbuild → test lab → certification
Scenario · parallel test lab
P80 inside launchOct 22
- P50 finishOct 14
- P80 finishOct 22 · 13 days earlier
- Certification criticality41% of runs
- Critical pathtest lab leaves the path
Foreweave highlights exactly which tasks enter or leave the critical path between scenarios — so the argument for the second test lab writes itself.
Plant the first task
The strongest future is the one you saw coming.
Import your current plan or grow one from scratch. Free while in preview — your first simulation takes minutes, not a consulting engagement.