PULSE — moving fast on the build without leaving a mess

PULSE turns the weekly by-hand schedule comparison into one repeatable, trustworthy analysis

Picture the person whose job, every week, is to figure out whether a major project is still on schedule. The plan lives in Primavera P6, but P6 doesn’t do the comparison — so they export the original baseline and the latest update, open a spreadsheet, and line the two up by hand. What slipped? What’s at risk? Are we still tracking to plan? The answer depends on who’s doing the work that week, the exports are hostile to begin with (inconsistent dates, status crammed into fields, different shapes between files), and once it’s done there’s no clean record of how it was done. Multiply that across a portfolio and it stops scaling entirely.

PULSE takes that whole exercise and makes it a button. Upload the baseline and an update; it cleanses and normalizes the exports, builds comparable timescales, and produces a metrics view you can trust — the same way every time, with a history you can go back to.

A note on how it’s built: the team uses AI tooling to shorten the distance from a decision to reviewed, tested code. The architecture, the review, and the tests stay exactly as rigorous as they’d be without it — that’s how the timeline stays short without leaving a mess behind. PULSE itself has no AI features; the AI story is about delivery speed, not the product.

What is being built

PULSE is three layers that each do one job:

Layer	What it does
Web app	Projects, uploads, run analysis, reports, and the planning workspace
Pipeline service	Runs the ordered processing steps end to end
Data engine	Cleansing, timescales, and metrics

On top of the analysis sits the forward-looking half: an interactive pull-planning workspace where teams commit the next one to three weeks, record a variance reason the moment something slips, and lock the plan when it’s agreed. A reports framework (weekly work plan, look-aheads, flagged activities) and five-role access control round it out.

Decisions worth defending

The pipeline only reports success when nothing failed. Failures are loud — never swallowed and shown as a green checkmark. This removes the worst outcome in this domain: a quietly wrong analysis that someone trusts to make a schedule decision.

Every project gets its own isolated database. There’s no query path that can pull one project’s data into another’s view. The trade is that changes have to be applied across every project’s database — a cost we accept to remove any chance of cross-project contamination.

The engine reads each project’s own settings instead of hardcoding them. One codebase serves projects with different calendars and configurations, which removes the slow death of a separate fork per client that nobody can keep current.

Where it stands

PULSE is in active development and we’d rather be straight about it. The pipeline, the analytics, and the “Update Plan” Gantt surface are working. Reports and roles landed recently. The Kanban tab is stubbed, and the move from the command-line tooling to the always-on service isn’t finished. We don’t have adoption or performance numbers to share yet — when we do, they’ll be real ones.