Cascade raises $3.5M to help construction firms find and win projects
THE SO WHAT
Vertical software is moving up the stack from workflow to revenue engine—helping contractors originate and qualify deals, not just manage them. If you sell into construction or other low-software-penetration trades, assume the next wave of tools will be judged on bid volume and win rate, not UI polish.
READ THE SOURCE
MORE FROM THE WIRE
Startups & VentureLondon-based Arrakis, which makes AI agents for industrial companies, emerges from stealth with a $7.5M seed and a $30M Series A at a $140M post-money valuation
Capital is starting to bet that agentic AI will land first in industrial workflows, not office chat—Arrakis raising $37.5M in seven months on an “AI operating system” for factories is a clear example. If you run OT-heavy operations, expect vendor pitches to shift from dashboards to closed-loop agents that touch real equipment and safety-critical processes.
Startups & VenturePassionfroot raises $15M to expand its B2B creator marketplace to the US
A $15M Series A for a B2B creator–brand marketplace is another bet that “creators as B2B media channels” will formalize into its own supply chain. If you run marketing or partnerships, expect more structured platforms mediating these deals—and start instrumenting performance data so you’re not flying blind as budgets shift from traditional B2B channels.
Startups & VentureLed By DeepSeek, 10 Frontier Labs Rush Onto The Crunchbase Unicorn Board In June
Ten AI labs adding $65 billion of value in a single month is a reminder that capital is still flooding into frontier model bets, not just application layers. If you’re building on top, plan for a noisy vendor landscape and negotiate flexibility—model and infra switching will be a strategic asset.
Startups & VentureDatabricks’ cofounder just raised $20M to be the Switzerland of AI compute
A Databricks-adjacent team raising $20 million to be the “Switzerland of AI compute” is a bet that orchestration across heterogeneous clusters—not just more GPUs—will be the real control point. If you’re scaling training or inference, start designing for multi-cloud and specialized hardware now rather than locking into a single vendor topology.