The submittal does not move in your project system. It runs in threads and desks.
The RFI packet chased across three inboxes before anyone answers it. The insurance certificate re-attached by hand because no one can find the last one. The change order that lives in a thread until someone remembers to log it. Below is what that work looks like today, what it looks like once Silow has mapped it, and what it gives back.
How the work looks today. How it looks after.
Seven processes, in the order we would take them. Rows tagged "Measured" carry a number that came out of recorded work — read the note below. Rows tagged "Observed in the trace" name the mechanism as it runs without attaching a borrowed figure.
A submittal needs the spec section, the product data, the shop drawing and the prior correspondence assembled into one packet before it can go to the architect or engineer. Someone rebuilds that packet from scratch, hunts for the one attachment that is not where it should be, and the RFI clock starts the day it finally goes out — not the day the question was first asked.
The packet assembles from what already exists: the spec, the last submission, the attachments already on file. A human reviews and sends. The clock starts closer to when the question was actually raised.
Work that needs a second human — a sign-off, a review, a reply from the architect or the owner's rep — waited a median of about seven days in the teams we recorded. Work one person could finish alone waited about one. Roughly two thirds of all waiting sat on the far side of a handoff. On a job with a submittal log a page long, that is most of your float, quietly spent on nobody in particular.
The handoff becomes an object rather than an assumption. Nothing sits silently, because "waiting on someone" is a state the work is in, not a thing someone remembers to chase on a Friday.
In a recorded team, one document was manually located and re-attached across two messengers on five separate occasions in three weeks, because no shared link for it existed. On a job that document is a COI, a W-9, an insurance cert or a lien waiver — and the request for it is retyped from scratch every time by someone who already knows exactly what it will say and which sub is late again.
Standard requests draft themselves from the language your team already uses, with the right attachment already attached. The human reviews and sends. The file has one retrievable place, so it is never hunted for again.
A change order needs the original scope, the cost breakdown, the backup documentation and the prior correspondence pulled together into one packet before the owner or architect will even look at it. That pull happens by hand, under time pressure, usually while the field is already asking why the work has not started.
The packet drafts itself from the scope, the cost detail and the correspondence already on file. A human checks the numbers and the framing before it goes out. Nothing about pricing or scope gets decided by the system.
A task is moving about 3% of the time it exists in the teams we recorded — the other ~97% it sits in someone's queue. A submittal, a change order or a pay-app approval crossing the PM, the architect and the owner's rep lives in exactly that queue, and it is invisible to every tool you own.
The queue becomes visible: what is waiting, on whom, since when, and what would unblock it. You are not making anyone faster. You are attacking the ~97%.
Schedule slippage, open RFIs, submittal status, budget-to-date — pulled from the schedule tool, the project management system and a spreadsheet into one deck, on a cadence, by hand, by someone senior enough to know what actually matters this week.
The recurring pull-assemble-format cycle is drafted from the systems that already hold the numbers. The judgement about what the numbers mean, and what to tell the owner, stays where it was.
Counting AI by application name undercounted every person in a team we recorded, and one of them by around seven times — their assistant ran inside another tool, so the window reported the host's name, not the assistant's. If you bought licences for your PMs and your dashboard says nobody uses them, the dashboard is the thing most likely to be wrong.
Adoption measured from the work itself rather than from seats and logins — and, more usefully, the reason. An assistant nobody opens is almost never a model problem. It is that nobody mapped it onto a procedure anyone actually performs.
Read this before you trust a number above. The measured rows come from anonymized recorded teams in document-heavy operations — a distributed team of under ten people over five weeks, and a cross-border operations team. The numbers apply to this work shape: documents crossing desks, handoffs that stall, packets assembled by hand. Figures are rounded, and every one was computed before a model was allowed to write a sentence about it.
What an agent may touch, and what it may not.
In a business where a signature carries scope, cost and safety, this line is the product. So we draw it before anything is built.
- Assembling the packet: submittals, RFIs and change orders collected, named to your convention, filed, and chased when missing.
- Drafting the request for the certificate, the W-9 or the waiver for the fortieth time this job.
- Watching the queue: which submittal, RFI or approval has been waiting, on whom, for how long.
- Producing the record of what moved, when and in what order — as a by-product, not a task.
- The status reporting ritual: the recurring pull from the schedule, the PM system and the sheet into one deck.
- Pre-filling a change-order or RFI packet from what already exists, ready for a human to check.
- Every change-order, cost and scope decision. Silow never approves, prices, or signs a change order.
- Any decision about safety or compliance on a site. That stays with the people licensed to make it.
- The exception nobody has seen before — which is most of what you hired a PM for.
- Every draft an agent produces is reviewed before it is sent. Nothing auto-sends to an owner, an architect or a subcontractor.
- Any decision about an employee. No ranking, no scoring, no automated judgement about a person.
Silow never approves a change order, never signs, and never decides safety or compliance for a site. It removes the assembly work around the decision and leaves the decision — and the accountable human who signs it — exactly where it was.
Time, money, and the people you already have.
The rule for this section: not one invented percentage. Every figure is either measured in a recording or supplied by you — hours multiplied by your own loaded cost. A saving you cannot defend to an owner is not a saving, it is a slide.
We do not quote you an industry percentage. Every candidate above gets scored in hours against your own recording and ranked by payback. The anchors we can already point at, from document-and-approval teams: a task moving about 3% of the time it exists, a median wait of around seven days each time work crosses a desk, and one document re-found and re-attached five times in three weeks.
The first is hours × your loaded cost — your rate, not a benchmark we invented. The second is the one nobody counts: the AI you already pay for and nobody opens. In a team we recorded, app-name telemetry undercounted AI use for every single person, so the licence decision was being made against a number that was simply wrong.
No new system to adopt, no data to move, no process change while we look — Silow runs on the tools the operation already uses, inside the perimeter you choose. The capacity comes back into the same team, doing the same job with the assembly work removed.
A task is moving about 3% of the time it exists. Making the PM faster optimises that 3%. Your schedule float lives in the other 97%.
From a procedure nobody has seen to a layer you own.
The agents are the visible part. The thing that makes them possible — and that keeps paying after they ship — is the record of how the operation actually works.
It is written down somewhere, and that document is wrong. The real procedure — the one performed at 09:00 on a Tuesday when a sub is late on a COI — has never been observed, which is why every automation attempt starts with a workshop and a guess.
The real steps, from the recording, with the evidence attached. In one of our own recordings all thirty supporting quotes were checked back against the raw scenes they came from, not against a summary of them. Ranked by impact, effort, risk and payback.
Packet assembly, the document chase, the status draft. Each arrives as a specification an engineer can build from: the observed steps, the evidence, and the lines it must not cross. Human review stays in the loop by design.
The agents are consumers of something more valuable: a private, structured record of how your operation actually works, living inside your perimeter. Onboarding, search, handover, and whatever you build next, on data only you own.
The office work nobody has a line item for.
These six categories are not our taxonomy — they came out of the report on an operations team we recorded, ranked from what that team was observed doing all day. They are not construction-specific, which is rather the point: this is where the day goes in any coordination business, before a single tool gets touched on site.
The recurring pull-assemble-format cycle across the schedule, the PM system and the sheet.
The chase: who owes what, who is blocked, what has gone quiet.
Submittal, RFI and change-order cases assembled into one reviewable object.
The pre-read and the post-meeting actions, drafted from what was actually said.
The approval and document packs, rebuilt from scratch for every new sub or owner request.
Onboarding, access requests, and the tool blockers that stall a whole day.
The part other vendors leave out.
Two of the hypotheses we most wanted to be true were killed by our own data. The vendor who tells you what their product cannot do is the one worth the second meeting.
The recording sees activity, never outcome. Whether the schedule held, whether the change order got approved on time — that is not in the data. Join one column of your outcomes to our trace and it becomes provable. Until then, any ROI figure quoted at you by anyone is a guess.
A duplicate-work detector: about 44 recoverable minutes across nine people over five weeks, so there is no product. A meeting-ROI report: three quarters of what the metric called "meeting time" had no room attached. Both were cut before they ever reached a customer.
Numbers are computed before the model is allowed to write a sentence about them, so no figure gets invented. But a true number can still be attached to the wrong claim. We have caught exactly that twice — both times with human eyes, not with a check.
Days for legal. Weeks for the roadmap.
Before any recording, we agree where it runs — including fully air-gapped, on-prem or inside your own VPC. No keystroke logging, no microphone, no message content, no scoring of anyone. Those are product constraints, not policy promises.
Silow runs on the team's existing machines and tools. No migration, no new system, no process change while we look — the point is to see the work as it actually is.
Scenes become tasks; tasks become the real procedure. Every opportunity scored by impact, effort, risk and payback against your own hours — confirmed against the recording, or killed by it.
A specification, not a suggestion: the observed procedure, the evidence, the scenes behind it, and the boundaries it must not cross.
What you do with the capacity is yours.
If the honest answer for your operation turns out to be "there is not enough here to be worth it", we would rather find that in a two-week recording than sell you a year of it. Book a call and we will show you what the map looks like on your own tools, and what it would rank first.