New Intel: Upwind Security is building out brand, AI platform, and threat research leadership in parallel. Usually precedes either a mega-round or a category-defining product launch within 12 months.
Upwind Security
|
Cloud Security
Series A
New Intel: Waabi has no sales team. They route 100% of revenue through Uber Freight and Volvo instead of building enterprise sales. Most AV startups burn a quarter of their Series C on a go-to-market org. Waabi is betting two partnerships.
Waabi
Autonomous Trucking
Series C
New Intel: Risotto started as an AI helpdesk inside Slack but is now automating who-gets-access-to-what across Okta and Google Groups. That is the same job SailPoint sells to the Fortune 500 for six figures, being delivered to Series B companies through a chat thread.
New Intel: SolveAI is sending engineers on-site with customers instead of selling purely self-serve AI tools. Slower to scale, but much harder to rip out once deployed.
SolveAI
Enterprise AI
New Intel: Runway is building a consumer mobile app with AI character calls, storyboard tools, ad workflows, and in-app purchases. That is a different business from its web platform.
Runway
Generative Video
Series D+
New Intel: Compa is putting more weight behind expansion than new-logo sales. That suggests the product is becoming core HR infrastructure inside existing enterprise accounts.
Compa
Compensation Intelligence
Series B
New Intel: Deepgram is working on voice models that skip text entirely, going straight from audio-in to audio-out. If it works, today's voice agent stack starts to look slow.
Deepgram
Voice AI / Speech Recognition
New Intel: Breaker is betting autonomous navigation can work with passive sensors alone. If it scales, competitors using expensive lidar payloads have a cost problem.
Breaker
Defense Robotics
Seed
New Intel: SolveAI is positioning between AI code tools and big consulting firms. The bet is that enterprises want production help, not another demo or another services army.
New Intel: SolveAI is building in Rust and training engineers on it internally. That is an unusual choice for AI tooling, and it points to durability over demo speed.
