.webp)
Startup Intelligence Database
Discover what the world's best startups are building before it's announced publicly.
Data Extraction
Speech AI
Model Evaluation
Prediction
Optimization
Opalite Health
Real-time AI medical interpretation in 150+ languages with EHR integration.
Medical interpretation costs $3-5 per minute through phone services, and wait times delay care. Opalite does real-time speech-to-speech interpretation in 150+ languages with automated clinical documentation, deployed across 10+ US states. Built by a physician-researcher with 200+ publications and an ex-Apple Intelligence/Siri engineer.
Data Extraction
Model Evaluation
Prediction
Optimization
Personalization
Origin
Designs regulatory DNA that outperforms DeepMind for safer, more precise gene therapies.
Gene therapies need precise regulatory DNA elements to control where and when genes express. Origin's Axis model outperforms DeepMind's AlphaGenome by 6.7% on activity prediction, with free API access for researchers and pharma partnerships forming.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Optimization
One Robot
Builds task-specific world models for training and testing robot policies in simulation.
Robotics sim-to-real transfer fails because generic simulators do not match the physics of specific tasks. One Robot builds task-specific world models learned from real robot data, and the founders built manipulation systems at Google, NASA JPL, Tesla, and McLaren. Early sales to Amazon's Physical AI Labs.
RAG
Data Extraction
Model Evaluation
Prediction
Optimization
Noetic
Automates regulatory research, documentation, and lab matching for hardware product certification.
Hardware product certification (FCC, CE, FDA, UL) takes months of consultant fees. Noetic maps applicable standards, drafts documentation, and matches products to testing labs in seconds. The wedge is narrow but the pain is acute for every hardware startup going to market.
Computer Vision
Model Evaluation
Optimization
Simulation
Personalization
Ndea
AGI lab building hybrid deep learning and program synthesis for autonomous scientific discovery.
Chollet (Keras creator, ARC-AGI inventor) and Knoop (Zapier co-founder) building hybrid deep learning and program synthesis for autonomous scientific discovery. The thesis is that current LLM scaling will plateau and learning efficiency through program synthesis is the path to AGI. ARC-AGI-3 launching as the next benchmark.
Data Extraction
Model Evaluation
Prediction
Fraud and Risk Scoring
Synthetic Data
Overdrive Health
AI-native medical billing automating the entire revenue cycle from eligibility to denials.
Medical billing at small practices leaks revenue at every step: missed eligibility, wrong codes, unworked denials. Overdrive automates the full cycle with AI agents, and the founder built production healthcare AI at EliseAI after doing quantitative engineering at Jane Street.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Optimization
OctaPulse
Automates fish grading and health monitoring with robotics and computer vision at 90-95% accuracy.
Fish grading is done by hand, one fish at a time. OctaPulse's delta robotics and computer vision inspect fish at 90-95% accuracy in under 30 seconds, with a pilot at North America's largest trout producer. CMU robotics pedigree with alumni from Tesla, NVIDIA, and ASML.
Data Extraction
Model Evaluation
Prediction
Optimization
Fraud and Risk Scoring
Orthogonal
Enables AI agents to autonomously discover, access, and pay for APIs through a unified marketplace.
AI agents need to discover and pay for APIs, but today every integration requires manual key management and billing setup. Orthogonal is building the marketplace and payment layer for agent-to-API commerce, with 20+ API partners and partnerships across 10+ YC companies. The founder built payments at Coinbase and billing at Vercel.
Data Extraction
Computer Vision
Model Evaluation
Fraud and Risk Scoring
Multimodal AI
o11
Automates document and spreadsheet workflows natively inside Microsoft Office and Google Workspace.
Microsoft Copilot for Office costs $30/user/month and works poorly. o11 builds native add-ins (not web wrappers) that use COM architecture for deeper integration with Excel, PowerPoint, and Word. Hundreds of companies already using it. Launched on Product Hunt as 'Microsoft Copilot, But It Actually Works.'
Data Extraction
Computer Vision
Simulation
Personalization
Recommendation Systems
Origami Robotics
Builds robotic hands with matched data-collection gloves for smooth human-to-robot transfer.
Most robotic hands and data-collection gloves have different kinematics, which means policy transfer is lossy. Origami co-designs both with identical sensing and mechanics, eliminating the embodiment gap. Already selling to Amazon's Physical AI Labs.
Data Extraction
Computer Vision
Personalization
Fraud and Risk Scoring
Anomaly Detection
Overshoot
Runs vision-language models on live video streams with sub-200ms latency via a simple API.
New model capabilities have made computer vision newly automatable. Overshoot is close to the workflow where buyers already spend.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Optimization
Mango Medical
Automates surgical planning for orthopedic procedures from CT scans using agentic AI.
Surgical planning for shoulder replacements takes hours of manual CT scan interpretation. Mango automates it in seconds via API, already past $500K ARR before demo day, with an 8-figure LOI from a leading orthopedic company. Only company in YC W26 pursuing FDA 510(k) clearance.
Data Extraction
Model Evaluation
Prediction
Optimization
Personalization
MochaCare
Manages home care agency operations with AI agents backed by human support for reliability.
Home care agencies run on phone calls and spreadsheets. MochaCare pairs AI agents with human backup for scheduling, hiring, and compliance, targeting the reliability bar that pure-AI solutions cannot clear in a sector where a missed shift means a vulnerable person goes without care.
Data Extraction
Model Evaluation
Prediction
Personalization
Fraud and Risk Scoring
MouseCat
Automates fraud investigations with AI agents that review data, generate rules, and close the loop.
The architect of AWS Bedrock's agent infrastructure teamed up with a Coinbase fraud ML engineer. MouseCat's agents run full fraud investigations, generate rules, backtest them, and close the loop into production. Most fraud tools detect; this one investigates.
RAG
Computer Vision
Speech AI
Model Evaluation
Fraud and Risk Scoring
Luel
Supplies rights-cleared multimodal training data from 3M+ contributors in days, not months.
AI labs need rights-cleared training data, and most marketplace options are either synthetic or legally questionable. Luel has 3M+ contributors, delivers custom multimodal datasets in days with full provenance chains, and is reportedly one of the fastest-growing startups in YC W26.