Hi HN, we’re Rishi and Sahil. We’ve developed Rudus (https://www.rudus.ai/), an AI-powered takeoff and estimation platform built for concrete subcontractors.

Takeoff is the process of measuring and quantifying materials from concrete plan sheets. Rudus identifies every concrete structure (footings, walls, columns, slabs), pulls in related details, and eliminates hours of manual quantity calculation. Here’s a demo: https://www.youtube.com/watch?v=PAMNDRWEdlI.

The problem: Concrete subcontractors are the backbone of every building, but their estimating workflow hasn't changed in 20 years. Right now, a senior estimator opens a PDF, manually traces every footing and grade beam, then hand-builds an Excel spreadsheet with 300+ line items- volumes, formwork, rebar by bar size with lap splices and development lengths. Bids can take weeks and even months. Most firms have just a few estimators, meaning they physically cannot bid on most of the work available to them.

The software incumbent in this trade hasn’t been updated since 2020. Beyond that, every AI takeoff tool on the market was built for GCs and treats concrete as one checkbox, rather than working around how concrete estimators actually price work. We’re building Rudus for this trade and only this trade.

We started this when Sahil took a construction management class and realized how the estimation workflows hadn't changed in decades. We started cold calling, walking into offices with donuts, showing up at job sites, and everyone told us the same thing: slow estimation is the biggest bottleneck in growing their business, but every new product they've tried has failed. We quickly realized that the reason those tools failed is a lack of trust and frequent errors causing later problems. Estimators stake million to billion dollar bids on these numbers, and they are clear that they won’t trade their workflow for a black box. We took a different approach: software that intelligently accelerates their current workflows rather than replacing it by forward deploying our product into their current estimation workflow.

When an estimator uploads their structural PDFs to Rudus, we auto-classify every sheet (foundation plans, section details, footing schedules, frame elevations) and route each to the right processing pipeline. Computer vision detects concrete elements across the drawing set and follows cross-references across sheets to resolve dimensions and detailing, catching elements that plan-only tools always miss. Each element gets expanded into full assembly line items: concrete, formwork, and rebar with all the calculations an estimator would normally do by hand. A typical foundation package goes from a handful of assemblies to 80-120 priced line items. The estimator reviews, overrides where needed, and exports straight into their existing workflow.

We have a couple key advantages in the AI estimation space. The first is our focus on concrete, a niche part of construction. No one else is building this for concrete subs because the sheets vary drastically from other subtrades. For this same reason, VLMs and other generic solutions don't work. Instead, proprietary computer vision models are required, relying on training from massive amounts of customer data. We run multiple different models trained directly on our customers' takeoffs, and every interaction from our customers with our models becomes a training example, allowing accuracy per client to sharpen with use.

Our second advantage is in our product methodology, as we’ve chosen to build a copilot, not a black box. Most AI takeoff platforms try to replace the estimator completely by autonomously producing quantities, but the quality of the outputs with current models is poor, so the takeoff gets redone by hand anyway. After 100+ hours sitting in rooms with structural concrete estimators and completing numerous takeoffs ourselves, we’ve built around their actual workflow. The estimator starts the takeoff, and Rudus extends the work across the sheet by finding similarities, following cross-references, and understanding callouts. The estimator stays in control of every accept, override, and edit. The result is faster takeoffs they can defend, not unreliable AI output they throw away.

We’d love to hear what you guys think about our demo video (https://www.youtube.com/watch?v=PAMNDRWEdlI) or your experiences building out computer vision models, or anything you think is relevant!