We rebuilt FPM from the ground up. New forensic engine, new analysis pipeline, new interface, and deep AI integration throughout. The new platform is in early access with a group of forensic analysts while we finish the last pieces. This post was first published in March and refreshed in October 2026. Here's what we built, why, and what's still to come.
Why a Ground-Up Rebuild
When we started FPM, the goal was straightforward: automate the comparison work that forensic schedule analysts do by hand. The original platform delivered on that promise for basic workflows, but as we pushed deeper into network analysis, concurrent delay measurement, and driving path tracing, we kept running into limitations of the original architecture.
Rather than bolt improvements onto a foundation that wasn't designed for them, we made the decision to rebuild. The new platform is engineered from day one around a proprietary forensic engine that gives us complete control over how changes are detected, how impacts are measured through the network, and how delays are attributed.
A Purpose-Built Forensic Engine
The heart of the new platform is an analysis engine built specifically for forensic schedule work. Scheduling tools like P6 help project managers plan forward. Our engine is designed to help analysts look backward with precision, measuring what changed, why it mattered, and how it moved the milestones. That difference in purpose shapes every design decision:
01
Independent schedule verification
Every schedule update is verified independently. No reliance on stored P6 dates. The math is the source of truth.
02
Cause-and-effect tracing
Every gap between activities is classified by cause: stall, constraint, predecessor delay, or real schedule slack. No judgment calls about why a successor started late.
03
Path-aware impact modeling
Every change between schedule updates is traced through the dependency network to determine its effect on milestones. Parallel paths are measured independently, giving true concurrent delay quantification.
04
Deterministic outputs
Same XER files in, same results out. Every time, regardless of who runs the analysis. The engine removes subjective choices from the measurement step entirely.
AI-Native from the Start
The new FPM isn't an app with AI features tacked on. AI is woven into the architecture. Our MCP server exposes 40 tools that let any compatible AI agent set a schedule up, run the analysis and query the results directly. That means forensic analysts can use Claude, GPT, Gemini, or any MCP-compatible assistant to navigate results conversationally:
# Ask your AI agent:
"What drove the 45-day delay to Milestone 12
between the March and June updates?"
# The agent calls FPM's MCP tools to:
- Pull the driving path at each update
- Identify which activities changed
- Quantify each change's impact on the milestone
- Trace concurrent delays on parallel paths
- Tell each driving path alone, with its but-for date
The tool computes. The agent navigates. The analyst decides. Three layers, each doing what it does best.
What's Coming
Here's a look at the capabilities in the new platform, from what's already working to what's still in development:
Proprietary forensic engine with full network path analysis
Every delay traced through the dependency network to its root cause and milestone-level effect
15-category change detection between schedule updates
Duration, progress, logic, constraints, calendar, structural adds/removes, and more
Path-aware impact attribution
Every change measured through the network to its milestone-level effect, with parallel paths measured independently
Milestone paths
Each driving path to a milestone told alone, with the date it puts the milestone at, the date without it, and the rank float to the next path, on the Impact Paths page
Concurrent delay matrix with cross-window attribution
Impact by category per analysis window, delay and mitigation side by side
What-if worlds and saved scenarios
Apply any set of a window's changes to the base, or remove it from the update, and read every milestone. Scenarios save the steps for replay
MCP server with 40 forensic analysis tools
Setup, delay analysis, milestone paths, what-if worlds, change tracking and WOET. Works with Claude Desktop, VS Code, and any MCP-compatible client
Agent-driven setup
An AI client creates the analysis windows, picks the milestones, starts the analysis and polls it to completion
WOET: Day-by-day execution timeline (As-Planned vs As-Built)
Contemporaneous-period analysis per AACE MIP 3.3, with the day-by-day earning timeline in the app
Full platform redesign
New interface built for forensic workflow: Gantt, network explorer, Path-Gantt navigator, Impact Paths and What-if worlds
Activity history across every schedule update
Complete evolution of any activity in one view: duration, progress, constraints, logic changes, and milestone impact over time
Schedule data quality and violation detection
Automated identification of P6 dates that violate network logic, with health indicators and violation classification
Responsible parties
Map activity code values to the parties that own them, so every impact can be read by party
Report generation pipeline
Export analysis results to presentation-ready reports with full audit trails
Request Early Access
We're working with a small group of forensic analysts and construction claims professionals during this build. If you'd like to be part of that group, or if you just want to know when the new platform launches, reach out. We'd especially like to hear from practitioners working active delay claims who can stress-test the analysis against real project data.