.webp)
Startup Intelligence Database
Discover what the world's best startups are building before it's announced publicly.
RAG
Data Extraction
Speech AI
Model Evaluation
Fraud and Risk Scoring
Moda
Monitors AI agents in production with real-time failure detection and conversation replay.
LangSmith and Arize monitor LLM outputs. Moda catches behavioral failures specific to agents: the agent that claims it completed a task but did not, the tool call that silently errors. Conversation replay plus one-click editing lets teams fix and regression-test in a single workflow.
Model Evaluation
Prediction
Optimization
Simulation
Synthetic Data
Mantis Biotechnology
Unifies biomedical device testing and clinical trials with AI-powered digital twins.
Medical device testing requires expensive physical trials. Mantis runs virtual clinical trials using physics-informed digital twins, reducing time and cost while generating synthetic data for regulatory submissions. The founder previously built a physics engine for robotic surgery at her prior YC company.
Computer Vision
Model Evaluation
Optimization
AI Infrastructure
Martini
Collaborative AI filmmaking platform orchestrating the best video generation models for directors.
Runway, Sora, and Kling each generate video well. None give directors the cinematic controls (camera, lens, motion) they need. Martini orchestrates all of them in one browser-based editor with real-time collaboration and NLE export. Over 200 films produced on the platform so far.
RAG
Data Extraction
Model Evaluation
Prediction
Anomaly Detection
Mendral
Always-on AI DevOps engineer that monitors, diagnoses, and fixes CI/CD pipeline failures.
The founders built Docker and Dagger. Now they are building an AI DevOps engineer that monitors CI/CD pipelines, diagnoses failures, and opens PRs with fixes. 16,000+ CI investigations autonomously closed per month across 15 teams. If anyone understands CI/CD at a structural level, it is these two.
Data Extraction
Speech AI
Model Evaluation
Optimization
Personalization
Maven
Enables AI voice agents to securely collect PCI-compliant payments during live phone calls.
Voice AI agents are handling millions of calls, but none of them can take a payment. Maven adds PCI-compliant card capture during live AI phone calls via a single API, with support for Stripe, Adyen, and Braintree. First infrastructure layer purpose-built for agent-to-human payment moments.
RAG
Data Extraction
Model Evaluation
Prediction
Optimization
Maywood
Automates the full M&A deal lifecycle from document generation to buyer analytics.
M&A deal execution still runs on manually assembled CIMs, pitch decks, and due diligence binders. Maywood auto-generates all of them from a unified workspace, learning each bank's style. The CEO ran PE deals at Blackstone, so the product is built from the buyer's side of the table.
Data Extraction
Computer Vision
Speech AI
Model Evaluation
Prediction
Menza
Acts as a 24/7 virtual data analyst connecting to 650+ data sources for plain-English analytics.
650+ data source integrations, 234% MRR growth since January 2026, and 100% customer retention from April 2025 onward. The co-founder exited ShortlyAI to Jasper and the CTO comes from Goldman Sachs quantitative finance. Menza is a virtual data analyst that actually sticks.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Optimization
Lexius
Detects shoplifting in real time from existing CCTV using privacy-first behavioral analytics.
Retail shrink costs $100B+ annually and existing loss prevention is either expensive hardware (EAS gates) or biased facial recognition. Lexius detects shoplifting behaviors from existing CCTV with zero new hardware and ignores personal characteristics entirely. Early pilots with 7-Eleven, Erewhon, and Prada.
RAG
Data Extraction
Model Evaluation
Prediction
Optimization
Lucent
Automatically finds bugs and UX friction by analyzing real user session replays with AI.
FullStory and Hotjar require humans to watch session replays. Lucent watches them automatically with AI, flagging bugs and UX friction across 30+ YC products. The founder already exited an AI company to Canva, and she is now selling the behavioral data back to frontier labs for training browser agents.
Data Extraction
Speech AI
Model Evaluation
Prediction
Optimization
Lance
Runs 24/7 autonomous AI agents that handle guest calls, sales, and operations for hotels.
Hotels cannot staff front desks 24/7, and every missed call is lost revenue. Lance deploys AI agents across phone, SMS, and email that handle guest communications, qualify sales leads, and coordinate operations. One clinic generated $350K in bookings from Lance-managed interactions in a single month.
Computer Vision
Model Evaluation
Prediction
Optimization
Fraud and Risk Scoring
Kyten Technologies
Designs custom aerospace-grade battery packs in days using software-defined manufacturing lines.
Aerospace battery packs take months to custom-design. Kyten delivers in days using software-defined manufacturing lines. Both founders shipped 6,000+ flight-qualified packs to orbit at SpaceX Starlink and Shield AI, and they are one of the few teams alive who have done this at scale.
RAG
Computer Vision
Speech AI
Model Evaluation
Prediction
Laurence
Optimizes Amazon PPC bids hourly using reinforcement learning and transformer models.
Amazon PPC is a $50B+ market where most sellers either overpay agencies or under-optimize with rule-based tools. Laurence recalculates bids hourly using reinforcement learning, already managing $10M in live ad spend with a 40% reduction in advertising cost of sales. The CEO built the Data Science Agent at Google.
Data Extraction
Model Evaluation
Personalization
Recommendation Systems
Natural Language Processing
LegalOS
Automates US employment-based immigration workflows with AI and attorney oversight.
Employment-based immigration filing has a 99.8% success rate when done right and devastating consequences when done wrong. LegalOS automates the paperwork with AI while keeping attorneys in the loop for judgment calls, targeting the H-1B/O-1/L-1 volume that BigLaw overcharges for.
RAG
Data Extraction
Computer Vision
Speech AI
Model Evaluation
Librar Labs
Gives the 98% of schools without a library system an AI-powered cataloging and search platform.
98% of schools worldwide lack a proper library system. Librar collapses cataloging from weeks to hours using camera-based bulk scanning and a self-healing data backend. The niche is small but uncontested, and the data asset (structured literary metadata) compounds.
Data Extraction
Computer Vision
Model Evaluation
Prediction
Fraud and Risk Scoring
Kita
Extracts decision-ready signals from messy borrower documents in emerging markets where OCR fails.
Emerging market lenders reject borrowers because OCR cannot read handwritten payslips and local bank statements. Kita's vision-language models extract structured data from 50+ document types where traditional tools fail. Live in Philippines and Indonesia, expanding to Mexico and US underserved segments. Built at Stanford.