AI engineering project predictor startup Cascade wins backing from a16z accelerator

Construction and engineering firms have a chronic problem: by the time a project shows up on a public tender board, half the good relationships that win the work are already locked in. Cascade, a London and New York-based startup, thinks it can fix that by reading the paper trail months earlier — bond filings, budget announcements, permits, capital plans, property transactions, earnings transcripts, meeting minutes — and flagging projects before they officially exist. Andreessen Horowitz's Speedrun accelerator just led a $3.5 million seed round to back that bet, with Ada Ventures, Blitzscaling Ventures, Indico Capital, shuckerVC, G2C Ventures and Snowball VC joining in.

The founders, Hannia Zia and Joana Ferreira, are both ex-Google, and they met at UnlikelyAI, the startup built by William Tunstall-Pedoe, the guy who helped create Amazon Alexa. That pedigree matters less than what they're claiming to have already done: Cascade says it's surfaced more than $10 billion in project opportunities for clients, including firms that build for JFK and LaGuardia airports. That's not a small number for a company this young, and it suggests the early signal-reading approach is finding real traction with an industry that still runs a lot of business development off phone calls and gut instinct.

Zia's framing of the problem is blunt: every firm she talks to can name a project it should have won but never even got the chance to bid on. That's the gap Cascade is chasing — not better proposals, but earlier visibility. In an industry where margins are thin and competition for public infrastructure work is fierce, knowing about a bond filing three months before a request for proposal drops could be worth more than any pitch deck.

The money will go toward deepening Cascade's network and pushing adoption faster, according to the company. That's a modest use-of-funds statement for a seed round, but it fits the product: this is a business built on data coverage and trust, not flashy features. The more signals it ingests and the more firms it proves right, the harder it becomes for a rival to catch up on breadth alone.