If you've ever spent three weeks waiting on an architect to answer a question about a detail that should've been on the plans in the first place, you already understand the problem Zero RFI is trying to solve.

Zero RFI is a new AI platform that claims it can catch the conflicts, ambiguities, and missing details in construction documents before they turn into Requests for Information. The company just secured backing from General Catalyst — one of the biggest venture capital firms in tech, the same people behind Stripe, Airbnb, and Snap. That's serious money betting that RFIs are a problem worth solving with artificial intelligence.

But here's the question every contractor should be asking: does this actually work, or is it another Silicon Valley pitch dressed up in a hard hat?

Let's break it down.

What RFIs Are (And Why They're Everyone's Least Favorite Part of the Job)

If you're a GC, you know RFIs like the back of your hand. If you're a sub who mostly deals with smaller residential work, you might not run into them as often — so here's the quick version.

An RFI — Request for Information — is a formal question from the contractor to the architect, engineer, or owner. It happens when something in the plans or specs doesn't add up. Maybe the structural drawings show a beam at one elevation and the architectural drawings show it somewhere else. Maybe the spec calls for a product that's been discontinued. Maybe a wall section just... stops, and nobody drew what happens next.

When you hit one of these, you can't just wing it. You submit an RFI, and then you wait. The architect reviews it, maybe kicks it to the engineer, and eventually you get an answer. Could take a few days. Could take a few weeks. Meanwhile, your crew is either standing around, working on something else, or — worst case — building something that'll need to be torn out later.

Here's what makes RFIs so brutal on the business side:

  • Volume: A typical commercial project generates somewhere between 400 and 800 RFIs. Large, complex projects can push well past 1,000. That's not a typo.
  • Cost per RFI: Industry estimates put the average cost at $500 to $1,500 per RFI when you factor in the labor to write it, the management overhead to track it, and the schedule delays while you wait. Some studies peg it even higher.
  • Schedule impact: Each RFI that sits unanswered is a potential domino. One unresolved question about a mechanical chase can hold up framing, which holds up drywall, which pushes your finish schedule. You know how this goes.
  • Finger-pointing: RFIs become exhibits in disputes and claims. "We told you about this on RFI #347, and you took 22 days to respond." They're the paper trail that lawyers love.

Add it up across a $10 million project with 600 RFIs, and you're looking at $300,000 to $900,000 in direct and indirect costs tied to the RFI process alone. That's real money coming straight off your margin.

What Zero RFI Claims to Do

Zero RFI's pitch is straightforward: feed their AI platform your construction documents — plans, specs, schedules, the works — and it'll flag the problems before you ever need to write an RFI.

Think of it like a really aggressive plan review. The AI reads through all the documents, cross-references them against each other, and identifies conflicts, ambiguities, and gaps. A structural detail that contradicts an architectural section? Flagged. A spec that references a standard that's been superseded? Flagged. A dimension string that doesn't add up? Flagged.

The company says their system can catch 60-80% of issues that would eventually become RFIs. If that number is even close to accurate, the implications are significant. On a project with 600 RFIs, catching even half of them during preconstruction could save weeks of delays and hundreds of thousands of dollars.

The platform is designed to work during the preconstruction phase — after you've got a set of documents but before the crew mobilizes. You upload the plans, let the AI chew on them, and get a report of potential issues to resolve upfront. The idea is that you can send all those questions to the design team in one organized batch instead of dripping them out one at a time over twelve months of construction.

How It Supposedly Works

Zero RFI hasn't published every detail of their technology, but based on what they've shared, the system uses large language models and computer vision to:

  • Read and interpret architectural, structural, mechanical, electrical, and plumbing drawings
  • Parse written specifications and cross-reference them against the drawings
  • Identify dimensional conflicts between disciplines (the classic "the duct doesn't fit in the ceiling" problem)
  • Flag specification inconsistencies — materials called out in one section that contradict another
  • Spot common omission patterns that experienced contractors know to look for

If this sounds similar to clash detection in BIM, you're not wrong. But Zero RFI is positioning themselves differently. BIM clash detection works with 3D models and catches physical conflicts — two objects trying to occupy the same space. Zero RFI claims to go further by reading the intent of the documents and catching logical conflicts, not just spatial ones. A spec that calls for fire-rated drywall but the detail shows standard assembly? That's a logic conflict BIM can't catch, but a sharp superintendent would.

The promise is that their AI acts like that sharp superintendent — except it doesn't get tired, doesn't miss things on page 847 of the spec book, and can process an entire document set in hours instead of weeks.

The General Catalyst Backing: What It Means

General Catalyst isn't some random angel investor writing $50K checks. They're a top-tier venture capital firm managing over $25 billion. When they back a company, it signals a few things.

First, they've done their homework. Firms at this level don't invest on a pitch deck alone. They talk to potential customers, hire domain experts, and stress-test the technology claims. They clearly believe the RFI problem is big enough — and the technology mature enough — to build a real business around.

Second, it means Zero RFI will have resources. Venture backing at this level means the company can hire experienced engineers, invest in training their AI on real construction documents, and actually bring a product to market — not just run a demo. Check our AI construction funding tracker and you'll see this is part of a broader wave of serious money flowing into construction AI.

Third — and this is the part contractors should pay attention to — it means the smart money thinks construction technology is still underserved. Despite all the noise about AI, construction remains one of the least digitized industries on the planet. Investors see that gap as an opportunity. More companies will follow Zero RFI into this space.

But VC backing also means something else: the company needs to grow fast to justify the investment. That pressure can lead to overpromising, premature launches, and sales teams pushing product that isn't ready. We've seen this movie before in construction tech. Keep your guard up.

How This Fits the Bigger Picture

Zero RFI isn't operating in a vacuum. AI is hitting the preconstruction phase from multiple angles right now.

Estimating tools are using AI to speed up takeoffs and catch scope gaps — something we covered in our guide to AI estimating and bidding. Plan review platforms are using computer vision to read drawings faster. BIM tools are getting smarter about automated clash detection. Even the big project management platforms like Procore and Autodesk are layering AI features into their existing products.

What's interesting about Zero RFI is how specific the problem is. They're not trying to be an everything platform. They've picked one painful, expensive, universally hated process — the RFI — and they're attacking it directly. That kind of focused approach tends to produce better results than the "AI for everything" platforms that try to boil the ocean.

The broader trend is clear: AI in construction is moving from "interesting idea" to "actual products you can buy." If you want a wider view of what's happening, our 2026 AI tools roundup covers the full landscape. But the preconstruction space — plan review, estimating, document analysis — is where AI is making the most credible progress right now. It's where the documents are digital, the patterns are repeatable, and the cost of mistakes is measurable.

The Realistic Take: Hype vs. Hope

Alright, let's be honest about what we know and what we don't.

What Could Actually Work

AI is genuinely good at pattern recognition across large document sets. A human reviewer might catch 70% of issues in a plan set — they're experienced, but they get tired, they have blind spots in disciplines outside their expertise, and nobody reads every word of a 500-page spec book. An AI that can cross-reference every page against every other page, every detail against every spec section, around the clock? That's a legitimate advantage.

The basic technology is proven. Large language models can read and interpret text. Computer vision can read drawings. The question isn't whether AI can analyze construction documents — it clearly can. The question is whether it can do it well enough to be useful in a real construction workflow where a wrong answer is worse than no answer.

What's Probably Oversold

That "60-80% of RFIs caught" number? Take it with a grain of salt the size of a cinder block. Here's why.

Not all RFIs come from document conflicts. A big chunk of RFIs on any project are generated by field conditions — things you discover when you open up a wall, hit unexpected soil, or realize the existing conditions don't match the as-builts. No amount of AI document review will catch those because the information doesn't exist in the plans.

Another chunk comes from owner-driven changes, value engineering, and scope clarifications that evolve during construction. Again, you can't predict those from the initial document set.

Then there's the nuance problem. Construction documents are full of conventions, regional standards, and implicit knowledge that experienced contractors understand but AI might not. When a note says "match existing," a veteran super knows to go look at the existing condition before ordering material. An AI might flag it as ambiguous — technically correct, but not necessarily useful.

We've written about what happens when AI goes wrong in construction contexts. The risk with a tool like Zero RFI isn't that it catches too few issues — it's that it catches too many false positives, buries the real problems in noise, and creates more work instead of less. If the AI flags 2,000 "potential issues" and your team has to manually review each one to separate the real conflicts from the false alarms, you haven't saved any time. You've just traded RFI management for AI output management.

The Trust Problem

Here's the deepest challenge Zero RFI faces, and it has nothing to do with technology. It's trust.

When a senior project manager reviews plans and says "this is going to be a problem in the field," the team trusts that judgment because it's backed by twenty years of building things. When an AI flags an issue, the team still has to verify it. That verification step doesn't go away just because a computer found it.

Over time, if the AI proves accurate, trust builds. But that takes projects — real projects with real results that contractors can point to and say "yeah, that thing caught a $200,000 problem on our hospital job." Zero RFI doesn't have that track record yet. They're asking contractors to trust them based on demos and investor backing. That's a big ask in an industry that learned a long time ago to trust what's built, not what's promised.

What You Should Actually Do Right Now

If you're a GC or a large sub running commercial projects with significant RFI volume, here's the practical advice.

Don't buy anything yet. Zero RFI is early. The product needs to prove itself on real projects with real documents and real results. Early adopters in construction tech often end up as beta testers paying full price. Let someone else find the bugs.

Do put it on your radar. Add Zero RFI to your watch list. Sign up for their updates. When they start publishing case studies with actual project data — not marketing projections — pay attention. If you want to track the broader space, our complete AI guide is a good starting point.

Tighten up your own preconstruction process. While you're waiting for AI to mature, there's plenty you can do right now. Dedicated plan review sessions, RFI logs from past projects analyzed for patterns, cross-discipline coordination meetings during bidding — these old-school methods still work. The contractors who'll benefit most from AI tools are the ones who already have a disciplined process. AI amplifies what you've got. If your preconstruction process is chaos, AI just gives you faster chaos.

Think about your RFI data. Start tracking your RFIs more carefully if you aren't already. How many per project? What categories? What's the average response time? What percentage come from document conflicts vs. field conditions? This data will help you evaluate whether a tool like Zero RFI would actually move the needle for your specific type of work. It's also the kind of analysis that helps you calculate the real ROI of AI tools when you're ready to invest.

Stay skeptical, stay curious. The construction industry has a long history of technology promises that didn't deliver. But it also has a history of eventually adopting tools that genuinely save time and money. AI document analysis is going to work — eventually. The question is when, and which specific product gets it right first.

The Bottom Line

Zero RFI is tackling a real problem with real money behind it. RFIs cost the construction industry billions every year, and the idea that AI could catch document conflicts before they become field problems is genuinely compelling. General Catalyst's backing means this isn't a garage project — it's a well-funded bet on a specific, painful inefficiency.

But "well-funded" and "works on your projects" are two very different things. The technology is promising. The claims are plausible. The track record is nonexistent. And in construction, track record is everything.

Watch this one closely. If Zero RFI delivers even half of what they're promising, it could genuinely change how preconstruction works for commercial contractors. But don't sign a contract until you've seen the receipts — real project data, real results, from contractors doing work like yours.

That's not pessimism. That's how contractors have always evaluated new tools: show me it works on a job site, not in a demo room. AI doesn't get a free pass on that standard.