Backlog is thinner in 2026. Owners are slower to commit. More GCs are chasing fewer qualified projects.
That combination puts VPs of Pre-Construction in a bind. The instinct is to bid more. But bidding more without a filter means spreading your estimating team across pursuits that don't fit — and burning hours you can't recover.
The GCs with the best win rates right now are not the ones chasing every opportunity. They're the ones making faster, sharper go/no-go calls — and protecting their team's capacity for the bids that actually convert.
AI is changing how those calls get made. Not by replacing judgment. By giving VPs better information faster.
Every pursuit your team touches costs money. A full bid effort on a commercial project runs 30 to 40 hours of estimating time before you ever submit a number. On a complex ICI job, it can run higher.
When you're wrong — when you bid a project you shouldn't have — those hours are gone. And your team missed something they could have been winning instead.
The go/no-go decision isn't an administrative step. It's a capital allocation decision. You're deciding where to spend a scarce, expensive resource: your estimating team's attention.
That's why the best pre-construction leaders treat it like a discipline, not a gut check.
Most GCs have some version of a go/no-go scorecard. It usually covers:
These criteria are sound. The problem is that they rely on information you don't always have — especially early in the pursuit.
You get a notification that an RFP dropped. You have 72 hours to decide whether to engage. The documents are 800 pages of specs, half a set of drawings, and a draft agreement that looks like it came from the owner's last project. You have one estimator available.
Traditional scorecards assume you can answer the criteria before you've read the documents. In practice, by the time you've read enough to score it accurately, you're already deep into the bid.
That's the gap AI closes.
AI doesn't replace the go/no-go scorecard. It feeds it — faster and with more precision than a manual document read can deliver.
Here's where AI creates real leverage across the pursuit filter:
When an RFP drops, the first question is: what are we walking into?
A VP or pre-construction manager can use Provision's Chat Agent to query the documents immediately. Ask it which division of the specs applies, what the liquidated damages clause says, whether there's a performance bond requirement, or what the owner's dispute resolution mechanism looks like. Cited answers in under 20 seconds.
With Provision's Risk Review, the same document set gets scanned against a pre-built checklist of known contract and spec risks. Every risk is cited to the exact clause, section, or page. The 80% reduction in review time means your pre-construction manager can complete a meaningful risk screen in a fraction of the time it used to take.
That early risk profile changes the go/no-go conversation. You're not scoring the pursuit on relationship and project type alone. You're scoring it against real document risk — and you have the evidence in hand.
Once you've decided to engage, the next question is: what does this job actually require?
Scope complexity is a legitimate go/no-go signal. A project with ambiguous scope — conflicting drawings, incomplete specs, missing trade callouts — carries higher bid risk and higher post-award risk.
Provision's Scope Agent generates a complete scope-of-work package from the construction documents in under 60 minutes. It reads drawings, specs, and contracts together — not in isolation. That matters because scope gaps usually live at the intersection of documents, not inside any single one.
A $300K lead-lined glass scope was omitted from a hospital imaging suite and absorbed by the GC under "readily inferable" language. A $400K roof cover board missed on a $50M project was only recovered through a relational concession from the sub. These weren't estimating failures. They were document failures — gaps between what the drawings implied and what the specs specified. (Source: The Scope Gap Playbook, Chapter 5.)
If Scope Agent surfaces significant ambiguity at scope assessment — conflicting callouts, missing trades, unresolved design coordination — that's a signal. Either the project carries more risk than the fee supports, or you need to price that risk in explicitly before you bid.
The final stage of the pursuit filter is capacity. Can your team actually execute this bid well, in the time available?
This is where AI creates a different kind of leverage. When Scope Agent generates a scope package from the documents, it compresses 30 to 40 hours of manual scope-writing into under 60 minutes. That's not just efficiency. It changes the capacity math.
A team that used to be able to run two active pursuits in parallel can now run three or four — without cutting quality. That changes how you think about go/no-go. You don't have to say no to a pursuit because you don't have the bandwidth. You say no because it doesn't fit. That's a better reason.
Here's how to structure the pursuit filter when you have AI working alongside your team.
As soon as documents are released, run them through Risk Review. Flag the top 10 contract and spec risks before your team spends a single hour on takeoff. If the risk profile is severe — aggressive indemnity, no cap on liability, onerous milestone structure — that's a go/no-go input, not a post-award problem.
Apply your existing scorecard. Client relationship. Project type. Competitive field. Margin potential. Capacity. Score it honestly.
But now you have one more dimension: document risk. Add it as a scored criterion alongside the rest. Weight it appropriately — on a negotiated GMP, document risk matters more than on a hard bid with a well-defined scope.
If the pursuit scores above threshold on the standard criteria, run the documents through Scope Agent before you commit full estimating resources. Look for:
If the scope is materially incomplete, you have a choice: price the risk explicitly, seek clarification before bid, or walk away. All three are valid. None of them require you to find out at buyout.
Document your go/no-go decision and the criteria that drove it. This is discipline, not bureaucracy. Over time, it tells you which signals actually predicted win rate and margin — and which ones you were overweighting.
The GCs that get better at bid selection do it by learning from their own data. AI helps you generate that data faster. But you have to capture it.
The firms that have built this workflow consistently make the same observation: the benefit isn't just speed. It's confidence.
A Senior PM at a Canadian ICI GC put it plainly: "If we could catch three scope gaps or three missed items on every scope of work, then this thing pays for itself."
That's not a technology claim. That's a margin claim. Three scope gaps caught at pursuit stage versus three scope gaps found at buyout — or worse, in the field — is a material difference in project outcome.
Provision has reviewed more than $100 billion in project value and processed over 100,000 documents across GC pre-construction workflows. The pattern holds: the earlier you identify document risk, the cheaper it is to address.
There's a version of AI adoption that makes the go/no-go problem worse, not better. It's the version where you use general-purpose AI tools — ChatGPT, Copilot — to skim documents faster, and then mistake speed for accuracy.
General AI tools aren't built for construction document ingestion. They don't read drawings alongside specs. They don't handle addenda that update scope mid-pursuit. They don't output structured, trade-by-trade scope packages that an estimator can actually use.
A fast but inaccurate risk screen doesn't help you make a better go/no-go call. It just gives you bad information faster.
Purpose-built tools for GC preconstruction ingest the full project set — drawings, specs, contracts, addenda — and link every finding back to its source. That's the standard your pursuit filter needs to meet.
A good go/no-go process doesn't end when you decide to bid. It sets up the bid for success.
When you've completed a document risk screen and a scope completeness assessment before committing full estimating resources, your team starts the bid with a clearer picture of where the risk is. They know which trades need scope clarification. They know which contract clauses need to be flagged for legal. They know which spec sections are ambiguous enough to require a subcontractor RFI before bid day.
That front-loaded intelligence is the difference between a reactive bid and a controlled one. The Scope Gap Playbook frames this as one of the eight habits of high-margin GCs: front-load the buyout conversations. The same principle applies at pursuit stage. Don't wait for bid day to find out what you don't know.
VPs who build this discipline into their pre-construction workflow report getting through pursuits faster — Provision users consistently complete bid packages at roughly 2x the pace of manual workflows — without sacrificing the accuracy that protects margin.
The firms that get the most out of AI-assisted go/no-go decisions treat it as a workflow, not a feature. That means:
The last point is harder than it sounds. Client relationship pressure is real. A long-standing owner with a difficult project is still a difficult project. AI gives you the evidence to have that conversation internally — clearly, early, and without it being personal.
If you're building or refining your pursuit process, the scope of work template from Provision gives your estimating team a consistent starting structure once the go decision is made.
A go/no-go decision is the formal evaluation a GC makes before committing estimating resources to a pursuit. It scores the opportunity against defined criteria — client fit, project type, competitive field, margin potential, and capacity — and results in a decision to bid or pass. Done well, it protects estimating capacity for the pursuits most likely to convert.
AI reads the project documents — drawings, specs, contracts, addenda — and surfaces risk signals faster than a manual review. A VP or pre-construction manager can get a contract risk profile in hours instead of days, and a scope completeness assessment before committing full estimating resources. That makes the scorecard decision sharper and earlier.
Key signals include aggressive indemnity language, uncapped liability, heavy liquidated damages, onerous milestone structures, "readily inferable" scope clauses, and incomplete or conflicting drawings. Each of these transfers risk to the GC in ways that may not be visible until after award — or until a dispute arises.
With Provision's Risk Review, the screening process takes a fraction of the time a manual review requires — Provision users see an 80% reduction in contract and spec review time. A meaningful risk profile on a standard commercial project can be complete within hours of documents being released.
Go/no-go is the pursuit-level decision: do we invest resources in this bid at all? Bid qualification happens after the go decision — it's the ongoing assessment of how to structure the bid, price risk, and allocate subcontractor coverage. AI helps with both, but the go/no-go filter is where the biggest capacity gains happen.
No. AI surfaces information faster. The judgment — weighing client relationship, strategic fit, risk tolerance, and team capacity — still belongs to the VP. What changes is that the judgment is now grounded in real document evidence, not in what the team had time to read before the decision window closed.
Incomplete documents at RFP stage are themselves a risk signal. If you can't assess scope completeness because the drawings aren't there yet, that's worth flagging in your go/no-go scorecard. AI tools like Scope Agent will tell you what's missing — which gives you the basis to seek clarification or price contingency before you commit.
Risk Review screens contract and spec risk in hours, not days. See it on your next RFP.
Book a demoMore Articles
