Reads the submission you do not have time to read
Drafts what changed between revisions and what it cost in days, with the red-flag patterns surfaced and the driving activities named.
Zepth Rx · Schedule Intelligence
Program & Schedule Assurance
Catch the silent slip the moment it is uploaded — every revision recalculated, compared, and the movers named.
Last updated
Zepth Rx module
Program
AI agent built into the module
The snapshot chainIts own critical path calculationRevision comparisonSlippage attribution
14 checks
the DCMA 14-point assessment — the industry’s standard schedule health check
US Defense Contract Management Agency; documented and critiqued in Ron Winter PSP, “DCMA 14-Point Schedule Assessment” (2011)
A standard, not a statistic — and an unofficial one. It came out of a 2005 US defence memo requiring an integrated master schedule on contracts above $20M, and its documentation is essentially an online training course. It became a standard because everybody started using it, which is worth knowing before you make it contractual.
25%
of projects came within 10% of their original deadlines
KPMG Global Construction Survey 2015, “Climbing the Curve” — owner survey, prior three years
Eleven years old, and we are saying so rather than passing it off as current. Three projects in four missed their deadline by more than a tenth, and the industry appears to have stopped publishing the measurement.
Overview
A monthly programme submission is the most information-dense document on a project and the least examined. It is typically reviewed by eye, against the previous one, by somebody with a fortnight of other work to do. Re-sequenced logic, a quietly shortened duration, a new constraint suppressing float, twenty inserted activities on the critical path — all of it is in the file, and almost none of it is seen.
Program reads it properly, every time, in full. Every import becomes a dated, locked revision; Zepth Rx runs its own critical path over the network the contractor submitted, compares it to the last submission activity by activity, and names the ones that moved the completion date.
Identity, not codes
Activity codes renumber between revisions. Insert twenty activities into a P6 programme and the outline numbers shift beneath everything downstream. Compare two revisions on activity code and you get nonsense — activities that appear to have been deleted, durations that appear to have changed, delay attributed to the wrong work.
So every activity carries a stable identity that survives renumbering, and every cross-revision comparison, every attribution and every roll-up runs on that identity rather than the code. That is the difference between a comparison that produces evidence and one that produces artefacts.
The telescoping rule
Slippage attribution has one arithmetic trap, and getting it wrong makes the whole output meaningless.
An activity’s contribution is its own finish shift minus the shift it inherited from the driving activity upstream of it. Without that subtraction, a single mid-path delay is counted again against every activity downstream, and the total bears no relation to the actual movement. With it, the contributions reconcile exactly to the terminal slip.
Contributions reconcile. The named drivers sum back to the completion-date movement, so the table can be checked rather than trusted.
Recomputed or as-typed. Each driver records whether its finish is a calculated date or one somebody entered. They are different kinds of evidence, and a forensic reader needs to know which they are looking at.
Why the date moved, decomposed
Beyond which activities moved, Program answers what kind of change moved the date — apportioning the movement across six causes: progress, durations, logic, constraints, calendars and scope.
It is computed over 64 re-runs of the critical path, every combination of the six, and combined using Shapley values. The reason is technical and it matters: attributing sequentially — change one thing, measure, change the next — gives a different answer depending on the order you pick. Shapley does not. Movement that no combination explains is published separately as unexplained days rather than distributed among the six to make the arithmetic look tidy.
Should I believe this recovery programme?
The question an owner actually has when a recovery programme lands. Program grades every duration cut in the new submission against four independent evidence sources — the trade’s own history, the planned resource, recorded manpower, and measured field productivity — proportionally, as observed over required. A source that is not available is reported as not assessable, never as a failure.
There is deliberately no credibility score out of 100. A single number would invite exactly the false precision the module exists to remove. What you get is a stated verdict with the findings named.
How it looks
The instruments this module produces, drawn on illustrative data so the method reads clearly.
Every submission is stored, dated and locked. The contract baseline stays pinned where variance is measured from; the approved revision advances monthly. Comparing consecutive snapshots is how the platform sees change.
Contributions telescope, so they sum back to the terminal slip. Each driving activity’s contribution is its own finish shift minus the shift it inherited from upstream — without that subtraction a single mid-path delay is counted again against everything downstream.
- Pass
- Fail
- Not assessed — excluded from the score
A check that cannot be answered is not scored as a pass. With no baseline loaded, the checks needing one are marked not assessed and excluded from the score — a missing baseline must never flatter the number.
Where the slack has gone. Activities bucketed by total float with the near-critical band highlighted, and the float remaining on four driving paths across successive revisions.
The value
Why it matters
A slip found this month is a recovery conversation. Found in six months, it is a claim.
The comparison record is the contemporaneous evidence a windows analysis needs — built continuously, it costs nothing at the point of dispute.
Schedule health and a credibility verdict give a defensible basis for rejecting a submission, with the specific activities named rather than an impression.
An independent recalculation of the contractor’s own network is the cheapest assurance available on a project, and almost nobody does it.
Capabilities
What you can do
01
The snapshot chain
Every import becomes a dated snapshot with its data date, locked on arrival. A submission is evidence and is not edited after the fact — adjustments happen on an explicit working copy that keeps the same activity identities.
02
Its own critical path calculation
A full forward and backward pass over the submitted network: all four relationship types with lags, six constraint types, per-activity named calendars, retained logic or progress override, and data-date discipline. Then it compares its result to the dates the submission asserts.
03
Revision comparison
A structured account of what changed across 33 change types — activities added and removed, logic re-sequenced, out-of-sequence progress, calendar and constraint changes, compressed durations, rewritten actuals, lag abuse, open ends, negative float — plus compound red flags where several co-occur.
04
Slippage attribution
Which activities on the driving path moved, and how many days each contributed to the movement of the completion date — telescoped, so the contributions sum back exactly to the terminal slip.
05
DCMA 14-point health
Every submission scored against the fourteen checks, naming the offending activities rather than counting them. A check that cannot be answered is marked not assessed and excluded from the score, so a missing baseline never flatters the number.
06
Float and where it went
Total float tracked across the revision chain: burn rate, erosion pattern and projected exhaustion, plus float sequestration — float suppressed by date constraints rather than by logic, which hides a path’s true criticality.
07
More than one contractor
A project holds multiple programmes, each with its own snapshot chain, consolidated into a master on stable identity. Cross-programme dependencies, interface detection between contractors, and a conflicts register by type and severity.
The workflow
How it actually runs
- 1
Import the submission
Primavera P6 `.xer` or Excel. It becomes a dated snapshot, locked on arrival, with its data date recorded.
- 2
Recalculate independently
A full forward and backward pass over the submitted network, producing early and late dates, total and free float, and the ordered critical path.
- 3
Compare against the last revision
Change types, red flags, critical-path and float movement, and the driving activities whose shift produced the slip.
- 4
Score the health and grade the credibility
DCMA 14-point with the offenders named, and every duration cut graded against the evidence available for it.
- 5
Decide
Accept, reject or query the submission — with the specific activities to raise, rather than an impression.
AI that does the work
How AI changes Program management.
Grades the recovery claim
Where a new submission shortens durations, it assembles the evidence for and against each cut and states which sources could not be assessed.
The arithmetic is not the model’s
Critical path, float, telescoping and the Shapley decomposition are computed by deterministic engines. The model narrates the result; every figure it states in prose is reconciled against the engine’s own value, and the calculated figures govern.
The engineer’s judgment stays in charge; the AI removes the latency and the blind spots.
Best practices
- Designate the contract baseline explicitly and move it only on a formal, audited re-baseline with the reason recorded. Conflating it with the current approved revision is how variance figures quietly lose their meaning.
- Import every submission, including the ones nobody intends to approve. The chain is the evidence, and a gap in it cannot be reconstructed later.
- Read the independent recalculation against the submitted dates before anything else. Where they diverge, the divergence is itself the finding.
- Treat a check that could not be assessed as unanswered, never as a pass. A health score inflated by absent data is worse than no score.
Dashboards & reporting
The revision register, a full interactive Gantt with critical-path highlighting and baseline overlay, the forensic comparison view, float history with projected exhaustion, and the multi-contractor coordination roll-up. The comparison produces a print-ready credibility report.
Live dashboards
Drill-down & filters
Export to Excel / PDF
FAQ
Common questions
Which programme files can I import?
Primavera P6 `.xer`, and Excel (`.xlsx` / `.xls`) with a guided column mapping. Microsoft Project users export to `.xlsx`.
Does Zepth Rx write a programme?
No. It reads the programme your contractor submits and gives you the instruments to interrogate it. The contractor authors; you assess.
What happens to a submission after it is imported?
It is locked. A submission is evidence and is not edited after the fact. If a revision needs adjusting you fork it into an explicit working copy that keeps the same activity identities, so the chain stays intact and the change is visible as a change.
How many revisions do I need?
One gives you CPM, float, DCMA health and the Gantt. Comparison, slippage attribution and credibility grading need two or more — they are all statements about change.
What if the contractor’s dates disagree with your recalculation?
That divergence is itself a finding, and it is reported as one. Zepth Rx does not take the submitted dates on trust; it runs its own forward and backward pass over the network and compares.
Can it handle several contractors on one development?
Yes. A project holds multiple programmes, each with its own snapshot chain, consolidated into a master on stable identity so an activity appearing in more than one programme is reconciled rather than double-counted.
Sources
- US Defense Contract Management Agency — the 14-point schedule assessment
- Ron Winter PSP — “DCMA 14-Point Schedule Assessment” (2011)
- KPMG Global Construction Survey 2015 — “Climbing the Curve”
Related modules
Zepth is the construction project delivery platform — it runs construction, procurement and asset management on one record, and does the work: reading the drawings, reviewing the submittals, matching the invoices and flagging the risks, with a human sign-off on anything consequential.
See it on your project.
A short, tailored walkthrough on your real workflow — no generic demo.