Prephase is a construction management system built to match the shape of the work, from first selection to final walkthrough, instead of asking the work to match the software.
The most preventable, margin-killing failures in construction happen during the handoff. On a typical project, a single piece of information passes through a chaotic chain of custody: from the scope language that defines the work, to the product selection that specifies it, to the procurement log that tracks it, to the budget that absorbs it, to the subcontractor standing at the rough-in.
This data gets forced into scattered email threads, static PDFs, and detached spreadsheets with no version control and no validation, and it degrades at every step until scope that was never pinned down becomes a change order, or the wrong finish gets installed because nobody caught it in time. Lead times slip past without anyone tracking them and budgets drift until someone notices too late, and by the time the schedule falls behind, nobody can say exactly why. At the end of it, a client who paid a premium gets handed a disorganized pile of loose manuals for the things now sitting in their home.
None of that is inevitable. It happens because the system asks too much of the people doing the actual work, so the record falls apart right when it matters most.
Prephase is built around a simple idea: software should follow the shape of work that already happens on a job, not the other way around. Each tool matches how the work already gets done, so entering data isn't a separate task pulled away from the job, it's just part of doing it. That's why the data holds up. Nobody's cutting corners on a form nobody asked them to fill out in the first place.
Each tool is a standalone field utility built for one specific handoff. Commitment and Schedule are fully independent, running on their own data with no connection to anything else. Track reads its manifest from Residential, one-way, so procurement stays tied to the same verified spec instead of a re-typed copy of it. All five live inside one shell under one login, and that's a friction decision, not a data one. A GC juggling five separate logins ends up guessing instead of checking, which is the same failure that made spreadsheets and email threads unreliable to begin with. The shell doesn't connect the tools. It just means nobody has to choose between doing it right and doing it fast.
Prephase Scope: scans raw scope of work documents for ambiguous language and open-ended responsibilities. It flags triggers like "as needed," "match existing," or "allowance," and hands the project manager the exact question needed to close the gap before work is assigned. Not AI. Deterministic and field-mapped, so the same input always flags the same risk.
Prephase Residential: replaces uncoordinated spec documents with a structured, cascading data-entry portal covering plumbing, appliances, lighting, electrical, finishes, and hardware. The library ships fully populated, and grows as new jobs call for new brands and finishes. From that record, Residential produces a printed field record and a mobile version for installers:
Prephase Track: reads the product manifest directly from Residential, letting teams track vendor lead times, order dates, and field status. If an upstream design change happens, Track quarantines the item and requires a fresh confirmation before it can proceed, so nothing gets silently ordered wrong.
Prephase Commitment: anchors the project to a macro budget, distributes allocations to each trade category, and gives the team a live ledger with plain-language variance flags separating internal budget offsets from client-directed changes.
Prephase Schedule: tracks ten fixed macro phases as one continuous road, sets the planned baseline against real-world progress, and requires a note for any date deviation, so there's a readable history of how and why the schedule moved.
Most construction software gets built by developers who spent a week interviewing contractors. Prephase came out of a decade of actual field management and capital program controls, and every tool reflects that, not a theory of how the work should go.
But the real difference isn't who built it. It's what the tool asks of the people using it.
Most tools in this category treat a user's time as free. They want narrative summaries nobody wants to write or read, and they bury the answer a crew needs behind a menu built for an office, not a job site. Some try to catch ambiguity with AI, which means the same document can get flagged differently depending on the model, the day, or the prompt. None of that looks like much on its own, but add it up across a crew and a year and you get a record nobody trusts, in a tool people stop opening.