What SchX is
A scheduling platform with its own engine.
SchX is a deterministic scheduling platform with its own calendar-aware engine, Arc. It ingests a schedule, builds an isolated workspace, analyzes it, schedules and saves it the way P6's F9 does, stages every change for human approval and exports a clean file back, beside Primavera P6 or on its own.
How a schedule moves through SchX
Four stages, from the file you have today to a clean file back in P6.
-
1
Bring it in
Import from P6, XER, P6 XML, MS Project or Excel. You get a full record of what came in, and the schedule gets its own workspace.
-
2
Understand it
Arc schedules it. Quality checks, float analysis and risk show you where the plan is strong and where it isn't.
-
3
Change it safely
You or the assistant propose changes. You preview them, approve them, and can undo any of them later.
-
4
Send it back
Export a clean file to P6 or MS Project, or share reports in PDF, Word and Excel.
It doesn't guess.
Dates and float come from the engine, never from a language model. Every value shows where it came from.
It fits how you work today.
Bring schedules in from P6, MS Project or Excel and send them back the same way.
It's made for teams.
Project security, live collaboration, approvals and a full audit trail.
Deterministic meets intelligence
Two halves, one boundary.
They are not competing approaches. Automation computes anything with a single defensible answer. Intelligence interprets, explains and proposes. Most of the trouble in planning tools comes from asking one of them to do the other's job, so SchX wires them together with a hard line between.
The deterministic half
AvailableWorks out anything with one right answer
- Scheduling engine (Arc). Forward and backward passes, float, driving logic, the longest path and multiple float paths. Float paths follow P6's numbering and order, with one known ordering difference on a single reference project.
- Analysis. DCMA 14-point, logic audit, loops, open ends, interfaces and update-to-update comparison.
- Governance. Typed change sets, pinned snapshots, validated transactions, rollback and checked exports.
- Schedule options. The project's real P6 scheduling options, in a dialog laid out like P6's.
- Labeling. Every value carries its source. A number without one is treated as a bug.
The intelligence half
AvailableExplains, interprets and suggests
- Reads the schedule only through the engine's tools, never from memory.
- Answers with the driving chain as well as the date, and cites activity IDs, fields and counts.
- Turns a quality report into a fix list, activity by activity.
- Drafts whole sections of plan from templates and stages them for review.
It can't write, approve, apply or export. That rule is enforced by the server, not by instructions to the model.
Suggestions come back as staged changes. A person decides what goes in.
The model is a component, not the product. You can swap it without touching the part that produces the numbers. That's why SchX keeps working when the models change.
The Arc engine
Same dates as P6 F9. Available
In network theory, an arc is the link between two nodes. In a schedule, the nodes are activities and the arcs are the relationships between them. Arc follows those links forwards and backwards to set every date, float value and critical path. It does the same for scenarios and Monte Carlo risk runs.
- Two engines, one answer. A Rust engine for speed and a reference engine beside it. They must agree on every field before any engine change is signed off.
- Locked between releases. Nobody can change the engine without passing the full data-set comparison and an owner's sign-off.
What Arc calculates
- CalendarsProject, activity and resource calendars, holidays and shift patterns
- LogicFinish-to-start, start-to-start, finish-to-finish and start-to-finish, with lags
- ConstraintsStart and finish constraints, mandatory dates and as-late-as-possible
- ProgressData date, actuals, remaining durations, retained logic and progress override
- FloatTotal and free float, the longest path and multiple float paths
- What-ifsScenarios and Monte Carlo risk runs on the same engine
100,000 activities scheduled in about 2.3 seconds on a 4 vCPU virtual server. Synthetic network, CPM pass only.
Same dates as P6 F9
On reference projects compared field by field with a native P6 F9, Arc gave the same early and late dates, total and free float, critical path and longest path on every activity.
No regressions
In regression testing at the last engine sign-off, Arc re-ran 249 project data sets with zero regressions. Any engine change has to pass that full re-run first.
Checked as it runs
When a schedule comes from P6, Arc compares its answer with the results P6 saved in the file and shows you that they agree, so you can move between the two with confidence.
How Arc processes a schedule
Eight steps from file to finished dates. Available
Arc follows the same scheduling rules P6 uses, including the project's own scheduling options. Every step is traceable, so you can always ask why a date is what it is.
- 1
Load the network
Activities, relationships, calendars, constraints, progress and the project's own scheduling options, read straight from the schedule.
- 2
Check the logic
Activities are put in dependency order and any loop is caught before a single date is calculated.
- 3
Apply calendars
Weekly work patterns, holidays and exceptions, inherited base calendars, and 24, 12 and 8 hour days. Time is rounded to the minute the same way P6 rounds it.
- 4
Forward pass
Early dates from the data date, through finish-to-start, start-to-start, finish-to-finish and start-to-finish links with their lags, all nine constraint types, retained logic or progress override, expected finishes and actual dates.
- 5
Backward pass
Late dates back from the project finish or a must-finish-by date. Open ends are handled the way P6 handles them, and negative float is kept, not hidden.
- 6
Level of effort
Level-of-effort activities take their dates from the work they're linked to.
- 7
Float and paths
Total float on the project's chosen basis, free float measured on the right calendar, critical activities, the longest path, and multiple float paths numbered and ordered the way P6 does it.
- 8
Trace and label
Every result keeps a trace of what drove it, is labeled as SchX-computed, and is compared with the results stored in the file.
Nothing extra to run
No sandbox.
No code interpreter.
No extra VM.
Arc is part of SchX itself. Everything it needs to calculate a schedule is already there, so there's nothing to provision, nothing to keep running on the side and nothing that can drift between runs.
The engine is part of the product
Arc is compiled into SchX and runs in the same service. There's nothing to provision, start up or keep warm before a schedule can be calculated.
No generated code
The assistant never writes scripts to do date arithmetic. It asks Arc, and Arc answers. The same schedule gives the same dates every time, to the minute.
Private by design
SchX deploys in your environment, your private cloud or your data center. Scheduling never depends on third-party compute or a public AI service. Deployment options
Labeled, not assumed
Every date and float value shows where it came from, so you always know what you're looking at.
Fast at scale
About 2.3 seconds for the CPM pass on a synthetic 100,000-activity network, on a 4 vCPU virtual server.
AI is optional
Scheduling, analysis, risk and reporting all work with the assistant switched off. Turn it on per workspace when you want it.