.webp)
Startup Intelligence Database
Discover what the world's best startups are building before it's announced publicly.
RAG
Model Evaluation
Prediction
Optimization
Personalization
Vela
Autonomous scheduling agent coordinating complex meetings across email, SMS, and WhatsApp.
Calendly handles simple 1:1 scheduling. Enterprise recruiters scheduling thousands of interviews per week across time zones, availability conflicts, and VIP priorities need something different. Vela coordinates across email, SMS, WhatsApp, and phone with autonomous follow-ups. $5.5M seed, already running paid pilots and yearly contracts.
Model Evaluation
Prediction
Anomaly Detection
Reinforcement Learning
Workflow Agent
Visibl Semiconductors
AI coordination layer for chip design that detects drift, proposes fixes, and surfaces tapeout risk.
Chip design teams lose weeks to undetected design drift between specs and RTL. Visibl monitors continuously, auto-triages issues, and proposes verified fixes. The CTO built silicon at Intel, Arm, and Microsoft. It sits on top of existing EDA tools rather than replacing them, which is the only realistic go-to-market in semiconductor.
Data Extraction
Model Evaluation
Prediction
Optimization
Anomaly Detection
Valence
Unified prediction markets terminal aggregating liquidity across Kalshi, Polymarket, and Crypto.com.
Prediction markets are fragmented across Kalshi, Polymarket, and Crypto.com with no unified order book. Valence aggregates all three with smart routing, and over 1B contracts have been traded through the platform at $200M+ monthly volume. Built by a former Balyasny Asset Management L/S equity trader.
Computer Vision
Model Evaluation
Prediction
Optimization
Simulation
Voltair
Autonomous drones that continuously inspect electrical grids to detect faults and prevent wildfires.
50% of US power flows through transformers at least 30 years old, and failing infrastructure causes wildfires. Voltair's autonomous drones inspect power lines with thermal, LiDAR, and RGB sensors, detecting faults that helicopter crews miss. CAD $1.83M pre-seed from Invest Nova Scotia.
RAG
Data Extraction
Computer Vision
Model Evaluation
Prediction
Vector Legal
AI law firm for startups pairing LLM drafting with an ex-YC lawyer and Ironclad engineer.
New model capabilities have made legal tech newly automatable. Vector Legal is close to the workflow where buyers already spend.
Computer Vision
Model Evaluation
Prediction
Fraud and Risk Scoring
Anomaly Detection
Verdex
AI digital audit platform using satellite imagery to automate insurance claim verification.
Crop insurance claim verification requires sending adjusters to physically walk fields. Verdex does it from satellite imagery with computer vision, already live with a major US insurer covering 100M+ acres (over 11% of US farmland). Expanding to property and energy insurance next.
RAG
Data Extraction
Model Evaluation
Prediction
Optimization
Ventura
AI teammates for distributors automating quoting and order entry from emails and faxes.
Industrial distributors process orders from emails, calls, faxes, and PDFs. Ventura's AI agents parse all of it and push clean structured orders into ERPs. The founder built a prototype at Harvard Business School for a $2B distributor, then dropped out to build the company. Also authored the open-source magentic LLM framework.
Data Extraction
Model Evaluation
Prediction
Optimization
Fraud and Risk Scoring
Velum Labs
Automated privacy-first data quality OS with ML anomaly detection and encryption.
Data quality tools like Monte Carlo monitor for anomalies. Velum derives semantic contracts from actual data traffic, traces lineage across any stack, and enforces integrity from ingestion to dashboard, with no manual rules. The CEO studied quantum computing at Stanford and physics at Harvard.
RAG
Data Extraction
Model Evaluation
Prediction
Optimization
The Token Company
Compression middleware cutting LLM costs by 66% while improving accuracy by +1.1%.
An 18-year-old Finnish National Physics Champion built a compression middleware that cuts LLM input costs 66% while improving accuracy by 1.1%. Pax Historia (193B tokens/month, 15th largest on OpenRouter) saw users prefer compressed outputs and a 5% lift in purchases. 100K tokens compressed in under 100ms.
Data Extraction
Speech AI
Model Evaluation
Prediction
Anomaly Detection
Unisson
AI agents automating B2B software implementation and onboarding as expert product users.
B2B software implementation is a $200B+ services market dominated by Accenture and Deloitte. Unisson's agents act as expert users inside client products, handling onboarding, configuration, and support at software speed instead of consulting speed. 5-minute deployment, usage-based pricing.
RAG
Data Extraction
Model Evaluation
Optimization
Personalization
Travo
AI-powered real estate analytics with instant valuations, comps, and workflow automation.
CoStar charges $30K+/year for commercial real estate data. Travo uses vector search and ML pipelines to deliver instant valuations, rental comps, and zoning data starting at $49/month. Self-serve PLG for the brokers and investors CoStar prices out.
Data Extraction
Computer Vision
Speech AI
Optimization
Reinforcement Learning
Talking Computers
Builds TC-1, a proactive AI coworker paired with every employee.
The long bet: pair every employee with a persistent, proactive AI coworker that communicates through Slack, Teams, and email. The founding engineer has low-level GPU expertise (cuDNN, TensorRT, Rust, ex-Modular). Whether the product is too ambitious for YC stage is the question.
Speech AI
Model Evaluation
Optimization
Fraud and Risk Scoring
Anomaly Detection
Terminal Use
Cloud-native platform for deploying and scaling background AI agents with git workflows.
Deploying AI agents to production requires cobbling together VMs, monitoring, and scaling logic. Terminal Use packages it into a Vercel-like platform with git-native workflows and model-agnostic orchestration. All three founders built agent infrastructure at Palantir.
Data Extraction
Speech AI
Model Evaluation
Personalization
Anomaly Detection
Tensol
Managed, proactive AI employees that automate sales, support, and ops across 40+ SaaS integrations.
40+ SaaS integrations, 10+ pre-built AI employee templates, dedicated VM isolation per customer. Tensol ships managed AI agents that proactively monitor and act without prompts, built on the open-source OpenClaw framework. The founder previously built workflow automation at Stacksync (YC), so the integration plumbing is familiar territory.
RAG
Data Extraction
Speech AI
Model Evaluation
Prediction
Tepali
AI-native operating system for medspas unifying marketing, scheduling, and charting.
Medspas use 5-7 separate tools for scheduling, charting, marketing, payments, and inventory. Tepali consolidates them into one AI-native OS with virtual phone agents and automated SOAP notes. The founder scaled operations at Rilla (vertical AI unicorn) and brings PJT Partners M&A experience.