.webp)
Startup Intelligence Database
Discover what the world's best startups are building before it's announced publicly.
Data Extraction
Computer Vision
Model Evaluation
Fraud and Risk Scoring
Anomaly Detection
Klaus AI
Deploys a secure, always-on personal AI assistant from OpenClaw in under 5 minutes, no setup needed.
OpenClaw went viral as an open-source personal AI assistant. Klaus provisions a dedicated, pre-configured cloud instance in under 3 minutes with security, integrations, and API keys already set up. Heroku for AI agents, built by an ex-Jane Street quant and an ex-Amazon engineer.
Data Extraction
Model Evaluation
Optimization
Anomaly Detection
Reinforcement Learning
Haladir
Automates COBOL modernization with formally verified AI that guarantees translated code is correct.
Trillions of dollars in financial infrastructure still runs on COBOL. Haladir translates it with LLMs and formally verifies every line with TLA+, meaning the output is provably correct. IBM's watsonx Code Assistant does not offer that guarantee.
Model Evaluation
Simulation
Fraud and Risk Scoring
Workflow Agent
Natural Language Processing
Hex Security
Autonomous AI pentesters that find and exploit vulnerabilities using 150+ security tools.
Autonomous pentesting with 150+ security tools orchestrated by LLM agents, tested against hundreds of YC portfolio companies as live targets. That feedback loop of real-world vulnerability data makes the agents better every week in a way static scanners cannot match.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Optimization
Inviscid AI
Creates digital twins of buildings and data centers to autonomously optimize energy and cooling.
Traditional CFD simulations for building energy optimization take hours. Inviscid's physics-informed neural networks run in real time, producing physically accurate results that pure ML approaches cannot guarantee. The founders have PhD-level expertise in the exact technique.
Computer Vision
Model Evaluation
Optimization
Simulation
Fraud and Risk Scoring
HLabs
Ships US-made plug-and-play robotics hardware with Python SDKs for AI-powered robot development.
Robotics startups waste months sourcing actuators, motors, and control boards from fragmented overseas suppliers. HLabs ships US-made, plug-and-play hardware with Python SDKs. The founder ran Mystic AI (YC W21) for 6 years, so he has done the hard part of hardware-software integration before.
RAG
Data Extraction
Speech AI
Model Evaluation
Prediction
IncidentFox
Autonomous AI SRE that triages and fixes production incidents using multi-agent orchestration.
Open-source SRE agent with a multi-strategy RAG system (RAPTOR + Knowledge Graphs) achieving 74% Recall@10, supporting 24+ LLM providers with BYO API keys. The open-source wedge builds trust with the exact SRE teams who will pay for the managed version.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Optimization
GrazeMate
Autonomous drones that herd cattle using reinforcement learning, replacing helicopters and horses.
Helicopter mustering costs ranchers $300-500/hour. GrazeMate's autonomous drones use reinforcement learning to herd cattle with stress-minimized patterns. The founder grew up mustering 6,000 head on his family's Australian station and was publishing robotics research with CSIRO at 15.
RAG
Computer Vision
Model Evaluation
Fraud and Risk Scoring
Workflow Agent
Jinba
Lets enterprises build secure AI workflows through chat interfaces in regulated industries.
Zapier and Make let anyone build automations. They do not let compliance teams approve those automations before business users run them. Jinba separates build (Flow) from run (App) with enterprise governance, 100+ connectors, and SOC 2 compliance for regulated Fortune 500 buyers.
Data Extraction
Computer Vision
Speech AI
Model Evaluation
Prediction
Ishiki Labs
Builds socially aware AI meeting assistants that know when to speak and when to stay silent.
Otter.ai and Fireflies record meetings. Ishiki's Fern joins as a participant that knows when to speak and when to stay silent, surfacing expert answers in real time. Built by engineers from Meta's AR/VR programs and Citadel Securities' real-time systems team.
Data Extraction
Computer Vision
Optimization
Simulation
Reinforcement Learning
GRU Space
Building the first lunar hotel by converting Moon regolith into construction materials with AI.
Building the first hotel on the Moon using in-situ resource utilization to convert regolith into construction materials. Extreme long-shot, but the 500x increase in space power demand by 2030 means someone will need habitats, and GRU is the only company trying.
Computer Vision
Model Evaluation
Prediction
Simulation
Synthetic Data
Human Archive
Captures 8,000 hours/day of multimodal human activity data to train the next generation of robots.
Robotics foundation models are data-starved. Human Archive has 50,000+ contributors wearing custom sensor rigs across homes, restaurants, hotels, and construction sites, capturing 8,000 hours/day of synchronized video, depth, and tactile data. Scale AI for embodied AI.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Optimization
Eos AI
Unifies hospital data and automates clinical workflows for 3x productivity and 37% more revenue.
Hospitals have more data than any other industry but use only 3% of it for care decisions. Eos unifies fragmented clinical and operational data into a single AI-powered operating system, claiming 3x administrative productivity and 37% revenue recovery.
Data Extraction
Speech AI
Model Evaluation
Prediction
Personalization
FullSeam
Deploys AI agents that log into ERPs and banks to autonomously execute accounting workflows.
The founders built and sold TaxProper to Opendoor for 8 figures. Now they are building AI agents that log into ERPs and banks to execute accounting workflows end-to-end. A second-time YC team with a proven exit, attacking a $50B+ market.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Optimization
Fort
Auto-detects 50+ exercises and tracks reps, form, and fatigue on-wrist without manual logging.
Whoop and Oura track recovery. No wearable auto-detects and counts strength training reps. Fort does, across 50+ exercises, built by five ex-Tesla engineers who know how to ship sensor hardware with on-device ML. The category is greenfield.
Computer Vision
Prediction
Simulation
Anomaly Detection
Synthetic Data
General Astronautics
Builds autonomous robots that perform lab tasks in microgravity without human crew.
In-orbit research is bottlenecked by crew availability, not science. General Astronautics removes the human from the loop with autonomous lab robots that operate in microgravity, built by a SpaceX Starlink hardware engineer and a CMU roboticist.