Part one made the case that a data center campus is one module, tiled, and that the only honest way to price it is to measure the module completely and prove that measurement. Proof is not a one-time act. The number an estimator signs at bid is the same number that should govern the pour, the change order, and the closeout. That is what a single record is for, and it is where this whole series has been pointing.
The estimate is a record, not a deliverable
The bid side is shipped and running today. AI Takeoff measures quantities off the drawing set, classifying the plans and sections, calibrating scale, and reading schedules that live on separate sheets against the whole set rather than one sheet at a time. Estimating prices those quantities across nine stages on your own history, sub bids leveled and gaps priced as labeled allowances, with your estimator signing every line.
On the graph, that work does not get filed and forgotten once the bid is out. The set lands in AI Takeoff, the estimate becomes the record, and that same record is built to carry the build phase forward to closeout. A number is measured once and then referenced, not re-entered across a bid sheet, a buyout log, and a closeout binder that slowly drift apart.
The team that measures once carries that measurement to handover. Everyone else re-keys it four times and hopes the copies still agree.
One graph across every phase
Multi-site programs live or die on consistency, and consistency is what one record buys. The graph that holds the bid is meant to hold every phase of every site, feasibility through build, sitting beside the Procore and Autodesk environments your delivery partners already run, so program leadership reads real status off the record itself instead of a reporting layer stacked on top. Nothing gets re-keyed between phases because there is only ever one entry to read.
The shipped bid-side products already run on this record, the measured takeoffs and priced estimates in use today. Alongside them, Enterprise Search finds every spec answer, submittal, and precedent across the tools your program already runs. That much exists now. The question this part is about is how far the same record is meant to reach.
A portal for every seat
Where this is headed is the full life of the building on that one record, from the feasibility read all the way to the money rail that settles it, with a seat on the record for everyone the project touches. The owner, the estimator, the design partner, and the field would each read the same entry from their own view, instead of four exports that already disagree by Friday. The owner who audits everything would get provenance that does not stop at the estimate but runs to the last handover.
Roadmap: the full lifecycle on a single signed record, feasibility through the money rail, with a portal for every seat at the table. This is the direction Ruh is building toward, not a shipped product.
What runs now, what is in preview
It is worth being exact about the line, because the honest version of this story has one. The bid-side products, measured takeoffs and priced estimates, are shipped and in use. The feasibility engine, generative site schemes iterated on the actual parcel, and drawing set generation from the approved model run in preview on the same project graph. Preview is not delivery. Those design-phase capabilities are being tested with teams now, not sold as finished, and the money rail that would close the loop from bid to paid is further out than that.
Built to be shaped
The preview exists for a reason. Data center teams that want to shape the lifecycle are exactly who it is for. The building type repeats, so what one campus proves the next one inherits, and a team that puts its first module on the graph now is helping decide what the record looks like by the time it reaches build and, eventually, the money rail.
Roadmap: one signed record that runs from the first pad to the last handover, closed by the money rail. Not built yet; being shaped now by the teams in preview.
This series started at a raw site and ends at running halls, one project graph measured at every step. The next move is not another chapter. It is the second campus, opened not on a blank set but on the record the first one left behind, so the repetition finally works for the owner instead of against them.