What GanttAI actually does, feature by feature.

GanttAI is one product with five things in it. They are meant to be used in order, but each stands on its own, and most people start with one and add the rest when the first has paid for itself.

1. Review: the file becomes a schedule

Whatever the GC sent — a PDF print, Microsoft Project (.MPP), Primavera (.XER, .XML) or Excel — opens as a live schedule you can search, filter and zoom. The messy PDFs are included; that is most of the point, because a PDF is what most subs are actually given.

Then it finds your work. On a schedule it has never seen, scope extraction catches 80–90% of your activities. Correct it where it is wrong and it learns that job: by the second revision it is at 99%+. You end up with your scope as a list, which is the thing the printout never gave you.

It does not need the GC's logic to be useful. Without logic you still get the live schedule, your scope, manloading and full comparisons. With logic you also get dependency questions, what your work is waiting on, and 36 schedule health checks.

2. Compare: what the revision did to you

Pick a baseline and compare any later revision against it. It reports changes across dates, durations, logic and sequence at better than 99% accuracy, and separates the decisions from the drift so you are reading ten lines instead of four hundred.

Compare as often as the GC reissues. Rev 1 to Rev 40 of the same job is one project, and comparisons inside it are unlimited — there is no run counter and no credit to spend.

3. Crew planning: the people behind the dates

Give it hours per scope and hours per person-day. It derives the crew size behind every activity from those totals against the schedule's durations, and draws the curve: your peak, your shape, and the activities whose duration implies a crew you do not have.

Because the crew is derived rather than typed, the plan re-plans itself when a revision lands. A fifteen-day activity becoming nine does not change the hours, so the crew changes — and you see it before you staff to it.

4. Portfolio overview: every trade, every job

One page for the whole book of work. Trade-level manpower added up from every live project, by day, week, month, quarter or year, against the crew sizes you actually have. Utilisation where you are fat, shortfall where you are not, and a real answer to "can we take this one on" before you bid it.

5. The Agent: ask instead of look

It has already read the revision. Ask in plain English — dates, crew, sequence, float, what your work is waiting on — and get an answer against the schedule you just loaded, not a generic one.

It also produces files. Ask for a three-week look-ahead and a ready-to-use Excel file comes back out of the chat. Same for summaries. No filtering, no rebuilding it by hand every Friday.

What comes out

Everything exports: Excel, Primavera P6, Microsoft Project and PDF. Crew plans, comparison analyses and resource analyses download. Comparison reports share with a link, so the person who needs to read one does not need an account.

What it costs to find out

Pricing is per project, from $250 a month, and a project is one GC schedule and all of its revisions. Inside it, reviews, comparisons, crew plans, look-aheads and exports are unlimited on every plan — no seats, no credits, no overage. Archive a job you have closed out and the slot comes back.

The trial is one project, two reviews and one comparison, with no card and nothing that charges automatically at the end.