Cascade, a startup building an AI platform that helps architecture, engineering, and construction firms find and win projects, has raised a $3.5 million seed round led by Andreessen Horowitz Speedrun, with participation from Ada Ventures and Snowball VC. TechCrunch reported the funding on July 22, 2026, following an initial press release on GlobeNewswire on July 21 and earlier coverage from Forbes on July 20.

Founded by Hannia Zia and Joana Ferreira and launched in 2025, Cascade operates at usecascade.ai. The platform collects data from U.S. government websites at every level — state, city, county, and federal — where agencies post construction contract opportunities. According to The Next Web, Cascade can spot multimillion-dollar projects months before any formal request for proposals goes out, giving firms a head start on preparing their bids.

What sets Cascade apart from existing tools is its predictive feature. The system looks at past bid competitions to guess which companies are likely to win future construction deals. When customers tell Cascade whether they won or lost a bid, the AI uses that information to get better at predicting over time. Think of it like a sports analyst who refines their picks after learning the outcomes of past games.

Cascade has already signed contracts with firms that built JFK Airport, LaGuardia Airport, Four Seasons hotels, and data centers. Ferreira shared details directly with TechCrunch, confirming the startup's progress with established construction companies. Cascade's main competitors, GovWin IQ and ConstructConnect, also track construction opportunities but do not offer the same kind of win-probability predictions.

The founders applied to the a16z Speedrun program in September 2025. Speedrun is Andreessen Horowitz's program for very early-stage startups, and it led the seed round. Tech.eu and FinSMEs also covered the round on July 21.

Cascade plans to use the seed money for marketing, industry events, and hiring engineers. The focus on hiring engineers suggests that building the technology, not growing a sales team, is the company's biggest challenge. Gathering and organizing data from hundreds of different government websites, each with its own format and update schedule, is a large and complex task.

The construction industry's bidding process is long, unclear, and heavily reliant on personal connections. Firms usually hear about projects through industry contacts, public notices, and existing relationships. A tool that finds projects early and estimates the odds of winning against specific competitors fills a real gap. Whether the predictions become accurate enough to actually change how firms decide which projects to pursue is the open question — and one that the feedback-loop design is built to answer over time, after the product is in use.

Looking at the bigger picture, AI tools applied to paperwork-heavy industries like government procurement and construction tend to follow a pattern: first someone solves the problem of collecting all the data, then the prediction layer follows once there is enough historical information to learn from. Cascade seems to be working on both at once, betting that early customers who contribute win-loss data will give the platform an edge that is hard for competitors to copy. The risk is that older companies like GovWin IQ, which already have years of procurement data stockpiled, could build similar prediction features. If Cascade's advantage holds, it will be because the AI improves faster, not because it has access to data that others cannot get.