Every P6 schedule,
checked and explained.
Datum reads a Primavera P6 schedule and tells you what's wrong with it. Upload an XER or connect a live P6 database, and get an itemized, exportable analysis: DCMA-14 checks, logic and float problems, delay between revisions, and cost. Every finding is in plain language and cited to the exact activities behind it, in your browser or over the API.
The analysis behind every finding
Itemized findings
Every DCMA-14 check plus logic, float, duration, and constraint anomalies, listed activity by activity, not hidden behind a single score.
Comparison and delay
Diff two revisions at the relationship level, surface the driving path, and attribute slippage to the activities that actually moved it.
Cost and earned value
Budget, actuals, and earned-value integrity checks over PROJCOST and resource loading, where the schedule carries cost.
Grounded AI narrative
A prioritized, plain-language assessment of the schedule's reliability, with every claim cited back to the finding and the row it came from.
A finding, cited
Negative Total Float, 12 activities. The worst is −23 days on A-1040 (Install curtain wall). The current logic cannot reach the data date without acceleration, and the driving path runs through the curtain-wall sequence.
negative-total-float:A-1040Delay by area · vs Update 11
Illustrative example. Every figure traces to the activity that produced it.
Built for everyone who works with the schedule.
Write the monthly narrative for update 12.
Spend less time on the monthly review and more time planning the work.
What moved the finish date since update 11?
Know what moved the finish date, and why, before the owner asks.
Is update 12 ready to accept?
Review every submission the same way, and spot problems before you accept.
Which activities drove the nine-day slip?
Trace each delay to its cause, with the P6 data behind it for your report.
What does the record show about the sequence change?
Schedule facts with the P6 record behind each one, and nothing the data cannot support.
Give me a one-paragraph status.
One paragraph per project, written from the schedule, not from memory.
Upload or connect
Drop an XER export or connect directly to your Primavera P6 database. Every revision is retained.
Establish the datum
Findings, comparison, cost, and quality checks run on read, the reference every deviation is measured against.
Export the record
A cited narrative and itemized findings, ready to export and defend, or pull straight from the API.
Everything the incumbents detect, without the desktop, the black box, or the wait.
The old way
- Desktop software, licensed per seat, on one analyst's machine
- Findings buried behind a single quality score
- AI that cannot show its work, if there is any at all
- Re-run the whole analysis by hand on every revision
- A PDF that is stale the moment the schedule updates
With Datum
- Web and API, in the browser or your pipeline, nothing to install
- Every finding itemized, filterable, and exportable
- An AI narrative cited to the exact activities behind it
- Lossless history, new checks apply to schedules loaded months ago
- Live analysis on every upload, defensible for the record
- Read-only by designDatum reads your schedule and never writes back to P6.
- Isolated per organizationEach customer's data is kept separate, visible only to the people invited to it.
- Encrypted in transit and at restUploads and stored schedules are encrypted end to end.
- Every revision keptEach upload is stored unchanged, so any update can be reopened and re-checked later.
What schedules does it work with?
Primavera P6. Upload an XER export, or connect directly to a live P6 database (Oracle or SQL Server). Every revision you load is retained, so later checks apply to earlier schedules too.
Do I need to install anything?
No. Datum runs in your browser, and everything it does is also available over an API, so you can pull analysis straight into your own tooling.
What does it actually check?
The DCMA 14-point assessment; logic, float, duration, and constraint problems; revision-to-revision delay and driving-path changes; cost and earned-value integrity; and quality scoring against NASA, PMI, and PASEG references.
How do I know the AI is not making things up?
The narrative only explains findings the analyzers already computed, and every statement cites the exact finding and activity behind it. It does not invent numbers.
Is my schedule data private?
Yes. Each organization's data is isolated and stored encrypted. Datum reads your schedule and never writes back to P6. Narratives are generated through AWS Bedrock, which does not use your content to train models.
Can I use it today?
We are onboarding a small group of design partners running real P6 schedules. Request access below and we will talk through fit and pricing.
See Datum on your next update.
Built for schedulers, delay analysts, GCs, owners, and their counsel. We are onboarding a small group of design partners running real P6 schedules. Tell us about yours and we will be in touch.
Request logged.
Thanks, your request is on the record. We will be in touch about design-partner access.