280 features
Someone ran 631 job evaluations and sent 68 applications through a command-line tool built on Claude Code, and I keep thinking about what that says about who job search software is actually built for.
The changelog is one of the most underrated surfaces in software, and FeatDrop wants to make it a community.
Finding the right person has always been the hard part. Lessie thinks the answer is just describing who you want in plain English and letting an agent do the rest.
An AI that detects your Kubernetes incident, finds the root cause, and opens the pull request before you've finished reading the alert.
The most annoying part of AI-assisted UI work isn't the code, it's explaining what you're looking at.
Someone built a leaderboard for AI token consumption and I am genuinely unsure whether to admire it or be alarmed by it.
A virtual Tamagotchi that your AI agents can feed and play with sounds like a joke until you realize it's actually a clever stress test for MCP tooling.
Every AI email tool promises to sound like you. Stamp is betting it can actually pull that off.
The bet isn't that Notion gets smarter. The bet is that your AI tools stop being amnesiac every time they open a new tab.
Every AI writing tool asks you to stop what you're doing and go somewhere else. Clico is building on the premise that this is the actual problem.
Everyone is building voice agents. Almost none of them sound like anything a human would want to talk to.
The global employer-of-record market has too many players and not enough great products, which makes PIO's bet on conversational payroll either smart positioning or a crowded swim lane.
The most annoying part of using an AI coding agent is babysitting it. Anthropic just decided to fix that, and the fix is either very smart or a little terrifying depending on how much you trust classifiers.
TypeScript 6.0 is technically a maintenance release. It's also the most consequential version they've ever shipped.
Every AI app builder promises to get you to production. Zoer.ai is betting that none of them actually do, because they all started in the wrong place.
Every note app promises to get out of your way. Novi Notes is the first one I've seen that means it architecturally.
A Chinese AI lab just launched a model that reportedly participated in its own evolution, and the demo is convincing enough that I can't just scroll past it.
Managing MCP servers across a dozen AI clients is exactly as painful as it sounds, and someone finally built the obvious fix.
Minicor builds self-healing desktop automation that uses AI computer-use agents to operate legacy software like EHRs and ERPs that have no API, running on Windows VMs with video replay and automatic adaptation when UIs change.
Ontora is an AI process mining platform that connects to your CRM, email, documents, and conversations to automatically map how work actually happens, identify bottlenecks, and generate actionable insights at a fraction of the cost and delay of management consultants.
Tsenta is a desktop app that finds jobs matching your skills, optimizes your resume for each application, and applies in parallel across 12+ platforms including Workday and Greenhouse, all running locally on your machine with real-time visibility.
Datost is an AI data analyst that lives in Slack, connects to your databases and data lakes, and lets anyone on the team query, debug, and analyze data right where they already work, without writing SQL or waiting for the data team.
Primitive is building the communication infrastructure that fully autonomous agents need to interact with the outside world, handling the email, messaging, and notification layers that agents cannot currently access on their own.
Harbor replaces legacy electronic data capture systems for clinical trials with an AI-powered platform that builds itself from your protocol in ten minutes, captures data from source documents automatically, and handles compliance out of the box.
Mendral is an AI DevOps engineer built by the people who wrote Docker's first lines of code, and it automatically diagnoses CI failures, catches flaky tests, and opens pull requests with fixes before your team even sees the red build.
Salus is an API that wraps around AI agents to inspect and block incorrect actions before they execute, preventing the kind of costly mistakes that evals and observability tools only catch after the damage is done.
Scheduling Wizard automates physician and resident scheduling for hospitals, replacing the grueling manual process that chief residents and program directors endure with software already trusted by Johns Hopkins and Mass General Brigham.
Superset is a free, open-source terminal app that orchestrates multiple AI coding agents in parallel, each running in isolated Git worktrees, so developers can delegate ten tasks at once without agents stepping on each other.
Building a short-form video feed used to require a team of 20 engineers and a year of development. ShortKit gives you the same capability as a drop-in SDK with ML-driven pre-fetching, enterprise-grade transcoding, and native ad integrations.
Sitefire identifies the content that makes AI agents like ChatGPT and Gemini recommend your brand, then synthesizes AI-optimized web pages ready to publish. Traditional SEO is not enough when your customers ask AI for answers.
Unisson builds AI subject-matter experts that learn any B2B product in twenty minutes and then handle customer deployments, implementation questions, and technical support across Slack, email, and text.
21st.dev started as the largest React component registry for AI apps with 1.4M developers. Now they are building a full SDK for deploying AI agents with built-in UI, memory, observability, and sandboxed execution.
Large electricity consumers like data centers and industrial operations overpay for power because they lack visibility into their consumption patterns and hedging options. Condor gives them an AI-powered operating system for energy strategy.
Most fraud tools stop at detection. MouseCat goes further, running end-to-end investigations, extracting features from unstructured data, and converting findings into production-ready rules. Founded by ex-AWS AI and ex-Coinbase engineers.
Vector Legal gives startups both an AI platform for legal operations and an actual lawyer. VectorOS handles AI drafting, health checks, and contract intelligence. The law firm handles everything else.
Byteport builds file transfer infrastructure for companies that move massive files. Their DART protocol runs 10x faster than TCP for uploads ranging from 1GB to 100TB, serving robotics, satellite, AI, and enterprise applications.
Sila is building a team messaging platform where AI agents are not add-ons but native participants. Custom agents with dedicated roles sit in your channels, talk, listen, and work alongside your team.
SpotPay gives users one account that works across 40+ countries for receiving money, paying locally, spending with a card, and earning on their balance. TPV growing 43% week over week.
BeeSafe deploys AI agents that engage with scammers directly, wasting their resources and uncovering the financial accounts used for fraud. Backed by YC, Obvious Ventures, and America's Seed Fund.
Carson combines deep prospect research, branded deck generation, and plain-English workflow automation in a desktop app built by ex-Palantir engineers who take data security as seriously as their former employer did.
Aircraft maintenance technicians spend more time on paperwork than on planes. Zymbly is an AI copilot that handles troubleshooting, parts lookup, and documentation with aviation-grade security.
CodeWisp turns plain English prompts into playable web games. 17,000 games created so far, 2,000 weekly active creators, and a growing community that looks more like Roblox than a code editor.
Workers' comp cases drown in paperwork. Docura Health uses AI to turn thousands of pages of medical records into compliant reports, chronologies, and impairment ratings in under five minutes.
Supply chains still run on spreadsheets and email. Pollinate automates procurement workflows, starting with three-way invoice matching, using AI agents that work on top of existing ERP systems.
Resonate is rebuilding messaging from scratch with AI at the center. The first product is a voice messaging app currently in private beta on iOS, with an agentic video platform in the works.
Maywood automatically generates finished, on-brand CIMs, pitch decks, financial models, and diligence responses for investment banking deal teams, with SOC 2 Type II security.
Medspas run on five different tools that don't talk to each other. Tepali replaces all of them with one AI-powered system that handles scheduling, payments, documentation, and lead conversion.
Voxel Energy builds energy-independent data centers that capture, store, and consume solar energy onsite using second-life EV batteries and a novel DC microgrid architecture, bypassing utility grid connection delays entirely.
Fort is a $289 wearable that automatically detects exercises, counts reps, and provides muscle-level feedback for strength training, filling the gap that Whoop, Garmin, and the major wearable brands have ignored.
Protent analyzes live video feeds to identify early escalation patterns and predictive threat signals, giving surveillance staff the immediate context they need to intervene before situations become dangerous.
Travo provides the most comprehensive real estate data set with instant access to rental comps, ownership, zoning, GIS data, and financial analysis for any property or parcel in the US.
Corvera deploys AI agents that handle order processing, demand forecasting, inventory management, and logistics for CPG brands, claiming to increase profits by up to 40%.
DAIVIN! builds tankless breathing systems that eliminate heavy oxygen tanks by extracting breathable air from water, starting with diving and expanding to aerospace and emergency response.
Remy AI builds warehouse robots that automate operationally intensive, high-dexterity tasks using a proprietary model that learns new tasks with less training data than standard approaches.
Seeing Systems builds inexpensive, modular strike drones with swarm AI coordination, designed for rapid upgrades and more than 2x reduction in lifecycle cost, in use with the UK Royal Marines and allied forces.
AutoSitu deploys AI agents across city planning departments to review site plans against zoning codes and building ordinances, completing reviews up to 90% faster and 80% cheaper than manual processes.
Captain provides a straightforward API for AI agents to search through unstructured files with 95% accuracy and citations, indexing data from S3, SharePoint, and Google Drive at petabyte scale.
Patientdesk deploys an AI receptionist that answers patient calls around the clock, verifies insurance in real time, books appointments, and collects payments, all integrated with dental practice management software.
Crow plugs an AI agent into existing applications so users can accomplish tasks through natural language chat instead of navigating complex menu structures and workflows.
Oximy deploys a lightweight desktop agent that monitors AI tool usage across an organization, tracking adoption, spend, and security risks across 3,500+ AI tools.
Rhizome AI indexes 2.5TB of FDA and regulatory databases so pharma and biotech teams can get cited answers to complex regulatory questions in seconds instead of days.
ARC Prize Foundation builds AI benchmarks that test whether models can genuinely acquire new skills, and their ARC-AGI benchmark has been used by every major AI lab to measure what their systems still cannot do.
Carrot Labs builds specialized LLMs for your business workflows, then continuously improves them against your success metrics, capturing proprietary know-how that gets more valuable over time.
End Close automatically reconciles transactions across payment processors, banks, and internal systems, using deterministic matching rules and AI agents that investigate exceptions.
HLabs manufactures actuators, electronics, and robotic primitives in Austin, Texas, so American robotics teams can stop building everything from scratch and start shipping robots.
Sequence Markets helps crypto traders save millions in hidden slippage by sourcing liquidity across CEX and DEX venues in a single workflow.
Synthetic Sciences is building the infrastructure for AI co-scientists, starting with an agent platform that delegates literature reviews, trains models on cloud GPUs, and generates publication-ready LaTeX.
ZeroSettle provides a mobile SDK that switches customers from App Store billing to direct billing, letting developers keep revenue that would otherwise go to Apple.
Luel runs a two-sided marketplace where content creators get paid for their videos, voice, and images while AI companies get rights-cleared training data.
Training robot policies in the real world is slow, expensive, and bottlenecked by hardware. One Robot builds simulations that look real and behave real so you can iterate without the robot.
Scout Out is an AI-powered project management platform built specifically for residential contractors, handling everything from bid to completion in one place.
Doomersion turns the worst habit of the 2020s into a language learning method. Watch TikToks in your target language, matched to your level, and get smarter while you waste time.
What if Rome never fell? What if the Soviet Union survived? Pax Historia turns alternate history speculation into a playable, AI-driven sandbox game.
Power utilities spend tens of billions inspecting their own equipment. Voltair is building the drone fleet to do it autonomously.
Every company deploying AI agents will eventually ask who has access to what. Agentic Fabriq is building the answer before the question becomes a crisis.
Cattle ranching has not fundamentally changed in centuries. GrazeMate just automated the most labor-intensive part of it with drones that herd on command.
Most AI health apps are chatbots with a stethoscope icon. Prana is trying to be a full-stack medical provider that catches the things your annual physical misses.
Corporate accounting is the last place most people would expect AI agents to show up. FullSeam thinks that is exactly why it is the best place to start.
Zendesk was built for consumers. Intercom was built for product-led growth. Pylon is built for the messy reality of B2B support, where your customers live in Slack channels and expect you to remember everything.
Tapeout is the most expensive deadline in technology. Visibl Semiconductors is building AI agents that keep chip design teams from missing it.
A corporate card with 5% back at Home Depot, AI that finds hidden tax credits, and compliance tooling that actually understands security deposits. Goldbridge is building the financial stack landlords never got.
Lessen processes 3 million work orders a year across 280,000 properties with a 4.8 out of 5 satisfaction rating. They are not a startup. They are a quiet giant in property maintenance, and the AI layer they are adding could make them dominant.
While React frameworks grab headlines and funding rounds, Quasar has been building a cross-platform development toolkit on Vue.js that ships web apps, mobile apps, desktop apps, and browser extensions from a single codebase. And it actually works.
Tax research is the worst part of accounting. Basil is building an AI that reads the tax code so your CPA does not have to spend three hours confirming what they already suspected.
Expense reports are the cockroaches of corporate life. Nobody likes them, nobody has killed them. Koda uses AI to auto-categorize spending, flag anomalies, and give startup founders a clear picture of where money goes without the spreadsheet misery.
Immigration compliance is a disaster of forms, deadlines, and edge cases. Mayflower plugs into HR systems to automate screening, classify visa pathways, and generate attorney-ready documents before the paralegal finishes their coffee.
Home renovations are expensive, unpredictable, and full of information asymmetry. Brickwise applies AI to property management and maintenance, giving landlords and property managers tools that automate the most painful parts of keeping buildings in shape.
Claybird automates outbound sales for commercial service contractors. AI agents scan permits and directories, identify target businesses, and send personalized emails and calls to decision-makers.
Cleft embeds feature flags directly in your codebase. No third-party SDK, no external dashboard, no vendor lock-in. Just flags that live where your code lives.
Property management is one of the most operationally painful businesses to run at scale. Keet automates the tenant communication, maintenance coordination, and leasing workflows that consume property managers' entire days.
LiveBench releases new benchmark questions every month so AI models cannot memorize the answers. Eighteen tasks across six categories, all with verifiable ground-truth answers. No LLM judges. No stale data.
Property appraisals take days, cost hundreds, and still rely on a human squinting at comparable sales. Automax.ai does it in under 20 minutes with LiDAR and computer vision, fully GSE compliant.
Medicare's remote care management programs are worth billions, but most providers can't afford the staff to run them. Kaigo Health uses AI phone calls to automate patient check-ins at scale, turning a labor-intensive compliance exercise into a revenue machine.
Kubernetes incident response is a nightmare of runbooks, alert fatigue, and 3 AM pages. Kestrel AI deploys agents that find root causes and generate production-ready fixes automatically.
Police and military radio is still a mess of garbled transmissions and manual note-taking. Lakonia transcribes, identifies speakers, and extracts actionable intelligence from radio traffic in real time. No cloud required.
Voice agents and chatbots need fast retrieval. Like, really fast. Moss built a Rust and WebAssembly search runtime that runs locally on any device and returns results in under 10 milliseconds. No database server, no network hop, no excuses. Grammarly and HubSpot are already customers.
Pharmacies are drowning in phone calls about prescription refills. Remedy is building AI agents to handle the repetitive operational work so pharmacists can do what they were trained for: patient care. It is a brutally simple pitch for a brutally underserved market.
Annual pentests are a compliance checkbox, not a security strategy. Veria Labs replaces them with AI agents that hack your application continuously, prove exploits are real, and suggest fixes. The founders are ranked number one in US competitive hacking, which is exactly the kind of credential that makes this believable.
Andrew Wyatt spent his early career at a logistics startup watching small fashion brands struggle with a supply chain that was never built for them. So he built one that was. CALA is now the operating system that unifies design, manufacturing, and fulfillment for thousands of brands -- and just became the first fashion company to get access to DALL-E's API.
Internal audit is a $15 billion industry built on spreadsheets and manual checklists. Denki wants to automate 99% of it with software and keep just 1% as services. That ratio tells you everything about their ambition.
AI agents are about to start spending money. Locus is building the payment rails that let them do it safely, with spending limits, escrow, and audit trails. This is infrastructure that does not exist yet.
Building ML models still requires a team of specialists, months of iteration, and a GPU budget that would make most startups cry. Thesis thinks AI agents can compress that entire process into something one researcher can do in an afternoon.
The no-code web app space is crowded. Fastshot thinks the real gap is mobile. Describe your app in plain English, get a production-ready React Native build for iOS and Android. The question is whether that gap stays open long enough.
Every AI assistant lives in a browser tab. Logical thinks that is the wrong model. Their desktop copilot watches what you are doing across apps and proactively helps without being asked. It is the Clippy pitch done right: context-aware, privacy-first, and actually useful. The question is whether people want an AI that is always watching.
GPU costs are the single biggest line item for most AI labs. SF Tensor claims it can cut those costs by up to 80% by automatically sourcing cheap GPUs across providers and optimizing kernels for any hardware. Three brothers, all CS graduates before turning 20, are betting that vendor lock-in is the real bottleneck in AI research.
Mosquitoes kill more humans than any other animal on Earth. The current solutions are chemical sprays that poison everything else too. Tornyol is building autonomous micro-drones that hunt mosquitoes individually, and they claim it is 100x cheaper than what exists.
Google is no longer the only search engine that matters. ChatGPT, Perplexity, Claude, and Gemini are sending real traffic to real websites. Bear tracks where your brand shows up in AI recommendations and helps you show up more.
You have had hundreds of AI conversations that disappeared into the void. Nessie imports them all, organizes them into searchable knowledge, and lets you share curated AI brains with others.
Your financial life is scattered across a dozen apps that do not talk to each other. Selfin wants to unify all of it into one AI-native banking experience that actually gives you personalized advice.
Monitoring tools tell you something broke. Interfere tells you what broke, why it broke, and how to fix it before anyone files a ticket. That is a meaningfully different product.
800 million people use ChatGPT every week. Most businesses have zero strategy for showing up in those answers. Mimos thinks that is about to change fast.
Behavioral health clinicians spend hours on documentation, utilization reviews, and compliance paperwork. Perspectives Health built AI agents that handle it automatically, and they are growing 25% week over week across 9 clinics.
Every sales engineer has experienced the empty-state demo disaster. Eden generates production-quality synthetic data on demand so your product looks like it has actual customers in it, without touching production databases or spending a week building seed scripts.
Food distributors still take orders by fax, phone, and email, then manually key them into legacy ERPs. Burnt drops AI agents into those systems and claims 97% of orders now flow end-to-end without a human touching them.
AI coding agents have full access to your file system, your secrets, and your network. ContextFort gives security teams a way to see exactly what those agents are doing, without trusting the agents to self-report.
Four founders from YC's Summer 2025 batch are building implantable microchips that track cattle biometrics in real time. The pitch: early disease detection, reproductive insights, and what they claim is the world's largest cattle health dataset.
GPU data centers are burning through power at unprecedented rates and most operators have no idea how much they are wasting. DeepAware AI uses reinforcement learning to optimize workload placement and cut energy waste by up to 30%.
Every piece of hardware that transmits a signal or touches a human body needs certification. Normal automates the process that currently takes months of back-and-forth with testing labs and regulatory bodies.
Most trucking dispatchers plan loads using gut feel and spreadsheets. Palace builds AI that maximizes revenue per truck by unifying driver, operations, and freight data into a single planning platform.
Traditional computer vision requires months of labeling data and training custom models. OnDeck uses vision language models to analyze video footage out of the box, no training needed, and their NeurIPS paper says it works better.
Doe connects to 40+ business tools and deploys AI agents that handle repetitive operations work. At $500 per user per month, the bet is that replacing tedious tasks is worth more than most SaaS subscriptions.
Getting a USD account in Egypt or the UAE means paperwork, minimums, and waiting. Munify lets you open one with a national ID in minutes, then send money home at rates that make Western Union look predatory.
Industrial distributors still process purchase orders by hand, reading PDFs and retyping line items into ERP systems. Comena built AI agents that extract order data from emails and documents and push it straight into your back office.
Teachers spend five to ten hours a week grading math assignments by hand. Frizzle processes entire class sets in minutes, reading handwritten student work and generating step-by-step feedback that actually helps kids learn.
Cloudflare Tunnels, Zscaler ZPA, and Ngrok all let you expose internal services to the internet. Pangolin does the same thing, built on WireGuard, fully self-hosted, with identity-based access controls and 19,000 GitHub stars to prove people want it.
Ashwini Asokan left Intel to build AI that put humans first. She and her neuroscientist husband built Vue.ai into a $57M-funded platform serving Tata, FedEx, and Meta. Then the foundation model wave arrived and changed the math entirely.
Tax preparation software has not meaningfully changed in twenty years. Magnetic is using AI to read scanned documents, handwritten notes, and messy spreadsheets, then enter the data directly into legacy tax software like UltraTax and Drake.
Topology optimization is one of the most compute-intensive tasks in mechanical engineering. Topological built a foundation model called UToP that does it nearly 2,000 times faster than conventional methods with less than 5% error.
Coding agents are getting good at writing code. They are terrible at deploying it safely. Proliferate gives them sandboxed environments, event triggers, and verification workflows so they can ship without human babysitting.
The prompt-to-app space has a dirty secret: most tools generate frontends that look great and backends that are held together with duct tape. VibeFlow is trying to fix that by making backend logic visual, editable, and actually understandable.
A bad dimension on an engineering drawing can cost a manufacturer millions in scrap and delays. F4 automates GD&T compliance checks using AI, built by engineers who have lived through the pain at Tesla and SpaceX.
Enterprise voice AI has been a graveyard of bad IVR menus and robotic chatbots. Freya is betting that multilingual, compliance-aware voice agents can finally replace the call center without making customers want to throw their phones.
Home services technicians need to diagnose problems, recommend solutions, and close sales on the spot. Most are good at one of those three things. NOSO LABS builds AI agents that ride along on every service call, helping techs diagnose faster, price accurately, and sell with confidence.
Your inbox is a to-do list that other people write for you. April is a voice AI assistant that reads your email, drafts replies, manages your calendar, and filters the noise. Hands-free. All through your voice.
Traditional insurance prices hardware risk using actuarial tables built for a world of known failure rates. Iron Grid uses physics-informed ML models and live telemetry to underwrite batteries, robots, and industrial systems that actuarial tables have never seen.
AI agents are shipping into production with no way to safely test them against real-world failures. Playgent builds high-fidelity sandbox environments where teams can reproduce bugs, run evaluations, and train agents through reinforcement learning, all without touching live systems.
End-to-end tests are brittle, slow to write, and nobody wants to maintain them. Lark replaces Cypress and Playwright scripts with plain English descriptions and uses AI to figure out the selectors, waits, and assertions.
Commercial real estate brokers still run their businesses on spreadsheets, phone calls, and gut instinct. PARES AI wants to give them an all-in-one platform with CRM, skip-tracing, AI underwriting, and automated marketing.
Most AI companies want bigger models on bigger servers. Stellon Labs is going the other direction, building frontier-quality models small enough to run on phones, wearables, and embedded devices. Their open-source KittenTTS hit 8,000 GitHub stars in two weeks.
Every AI coding tool targets software developers. SigmanticAI built one for chip designers, automating the verification workflow that eats 70% of the semiconductor development cycle. Siemens is already a partner, and the fine-tuned Verilog models are beating Cursor on accuracy benchmarks.
Every major AI model today requires a data center to run. DeepGrove is building technology to bring frontier-level intelligence to edge devices, from phones to embedded systems, without sacrificing capability. If they pull it off, the implications for privacy, latency, and accessibility are enormous.
AI coding agents are good at writing code but terrible at knowing what libraries actually do, what APIs have changed, and what documentation says right now. Nozomio indexes external context so your agents stop hallucinating about frameworks they last saw in training data.
Search is splitting in half. Half the queries still go to Google. The other half go to ChatGPT, Perplexity, and Claude. The Prompting Company is building the playbook for showing up in the AI half.
Enterprise software implementations fail at staggering rates. Scope creep, miscommunication, and lost context between discovery calls and delivery teams destroy margins and timelines. Datafruit built AI agents that capture every conversation, structure scope from day one, and use historical project data to make every engagement sharper. Three UC Berkeley and Georgia Tech engineers are going after the consulting industry's oldest problem.
Notion and Google Docs help you write things down. ChatGPT helps you ask questions. Opennote combines both into a notebook where an AI tutor named Galileo reads your notes and helps you actually learn from them. Over 55,000 students are already using it.
Insurance underwriting still runs on spreadsheets, email threads, and tribal knowledge locked in the heads of senior actuaries. Tesora is building AI-native software that encodes expert workflows into durable systems for MGAs and carriers. Two founders with McKinsey, private equity, and Google engineering backgrounds think they can do for insurance what Harvey did for law.
Three former Apple engineers who co-created SwiftUI built a tool that lets you describe an iPhone app in plain English and get production-ready Swift code you can ship to the App Store. No Xcode required. No Mac required. You build the app on your iPhone.
Every productivity app you use has its own silo of context about you, and none of them talk to each other. Epicenter stores everything in plain text and SQLite files you own, then lets local-first apps share that memory. It is the anti-cloud-lock-in bet, and the founder has 10,000 commits a year to back it up.
Browser automation breaks constantly because websites change. Notte built a full-stack browser infrastructure platform that lets AI agents run on the internet at production scale, with under 50ms latency, 1000+ concurrent sessions, and an agent success rate above 90 percent. Two MIT researchers who spent years studying why web automation fails think they finally fixed it.
Companies are deploying AI agents without any system for tracking what those agents do, whether they drift from their intended behavior, or whether they comply with evolving regulations. TectoAI built a governance platform that treats AI agents like employees: onboard them, assign roles, monitor performance, and flag compliance issues before regulators do. Two founders from Google and regulated industries think governance is the unsexy layer that makes enterprise AI adoption actually possible.
AI agents are calling APIs, querying databases, and executing transactions with long-lived API keys and zero oversight. Alter intercepts every tool call an AI agent makes and enforces parameter-level authorization, ephemeral credentials, and real-time guardrails. Two founders in New York are building the identity layer that the AI agent ecosystem forgot to build.
A delayed clinical trial costs pharma companies up to $312 million. The biggest bottleneck is patient recruitment: research sites spend hours reviewing medical records and conducting screening calls for every single participant. Phases built an AI agent named Polly that reads records, conducts voice interviews, and schedules patients automatically.
Healthcare organizations are deploying AI agents without any centralized way to manage them. Wedge wants to be the operating system that deploys, governs, and scales those agents across entire health systems. I think they picked the right problem at the right time.
Semiconductor manufacturing has been stuck in 2D for decades. Atum Works, founded by Caltech and NASA engineers, is building a 3D nano-printer that fabricates multi-material structures at 100nm resolution for roughly one-tenth the cost of traditional lithography. NVIDIA has already signed a letter of intent.
EMTs spend a staggering amount of time on paperwork after every call. CareSwift, built by a working NYC EMT and his technical co-founder, uses AI to guide crews through documentation in under three minutes while catching compliance errors in real time. The result: 80% less time writing reports, fewer insurance denials, and medics who can get back to saving lives.
AI code generation is shipping features faster than ever. It is also shipping bugs faster than ever. Synthetic Society deploys swarms of AI users that behave like real humans to find the UX flaws, broken flows, and edge cases that manual QA misses. Two MIT and Columbia dropouts are betting that testing needs to evolve as fast as the code it checks.
Fleet management software has been around for decades, but most of it still feels like a spreadsheet with GPS pins. Flott HQ is building an AI-native platform that unifies planning, tracking, invoicing, and payments for transport companies doing $200K in ARR and growing 45% month over month.
Financial auditing is one of the last white-collar professions where people still spend days manually cross-referencing spreadsheets. Moby Analytics lets auditors build and deploy AI agents using only prompts and domain expertise, cutting 80% of the grunt work while keeping everything traceable and auditable.
Most AI chatbots refuse to give you a straight answer. Opine built 100+ personas that actually have opinions, sourced from real YouTube creators and podcast hosts, and it turns out people want advice that sounds like it came from a human with actual convictions.
Bloomberg Terminal costs $24,000 a year and looks like it was designed in 1998. Scalar Field is building an AI-powered alternative with financial agents, and 800 paying traders putting up $74K in monthly revenue suggests they are onto something.
Salesforce doesn't know how to sell a pressure washing contract. Cohesive built a CRM that automates sales and marketing for the trades, and 350 businesses are already using it.
Trucking runs on phone calls, faxes, and spreadsheets. Hemut wants to replace all of it with AI agents that answer phones, process documents, source loads, and handle accounting for small fleet operators.
Designing optical systems still involves expensive software, manual iteration, and a lot of guesswork. Photonium wants to automate the entire stack from design to prototyping, and their team has the physics chops to pull it off.
Tax season is a grind of document chasing, client questions, and manual review. Combinely thinks AI can handle the boring parts so accountants can focus on the judgment calls.
Lab instruments generate mountains of data that nobody can actually use. Labric wants to turn that mess into something AI can work with, and the timing might finally be right.
Everyone knows RLHF makes models better. Almost nobody has the infrastructure to do it themselves. RunRL is betting that turning reinforcement learning into a managed service is the right abstraction.
Same-day delivery of transplant organs and jet engine parts runs on phone calls and spreadsheets. chrt thinks that is insane, and they are right.
nao is an open-source framework for building analytics agents that actually understand your data warehouse. Not another chatbot with a SQL generator bolted on.
Two Berkeley dropouts built autonomous agent runtimes that get better at their jobs without human intervention. Their first agent, Rex, already handles GEO for 400+ customers including Rippling and Airwallex.
With the average cost of bringing a new drug to market now exceeding two billion dollars, early-stage biotech companies are turning to marketplace procurement platforms to cut overhead and redirect resources where they matter most — the lab bench.
Scheduling is boring infrastructure until it isn't, and Cal.com is betting that AI agents need a calendar the same way they need a browser.
AI video generation is moving fast but the tools are fragmented. Prism bundles six models into one editor with a timeline, storyboards, and lip sync. The bet is that creators want a studio, not a prompt box.
The AI email client market is getting crowded, but the price war might be the only fight that actually matters.
Every website builder promises you a store. Runner AI promises it will keep optimizing that store after you walk away, and that's a very different claim.
AI video tools have been promising to replace your motion graphics budget for two years now. Knowlify, backed by YC, might be the first one that actually has a thesis worth taking seriously.
Penetration testing has always been expensive, slow, and bottlenecked by a global shortage of people who know how to do it. MindFort wants to replace the wait with agents that run for hours and find what humans miss.
Everyone's AI presentation tool makes slides fast. Chronicle wants to make them good, and that's a harder problem than it sounds.
Designers spend more time on presentations, feedback wrangling, and asset resizing than they do on actual design. Ideate is the first platform built specifically to fix that — and 4,900 designers on the waitlist agree it's overdue.
AI agents are shipping faster than anyone can test them. Janus is building simulation environments that catch the failures before they reach production. The pitch: trust, but verify.
My Apple Music library has three copies of Abbey Road, a solo Lennon track filed under seventeen different album names, and recommendations so broken they once suggested I listen to myself.
Code review tooling has been a solved problem on paper and a daily annoyance in practice for years, and a small Mac app wants to fix that without asking you to change how you work.
The pitch is seductive: just say what you want and the AI figures out the rest. Whether that holds up past the demo is a different question.
Gen Z is leaving broadcast social media and retreating into group chats. Text.ai thinks the group chat itself needs an upgrade, and they built Alfi to be the friend in the chat who actually remembers what everyone likes.
The dirty secret of most teams using OpenClaw right now is that it's one person's setup, running on one person's machine, and everyone else is either SSH-ing in awkwardly or just not using it.
Everyone's chasing the frontier. Google quietly built something for the people who have to pay the bill.
Somewhere in your org right now, an engineer has Claude querying your production database, and your security team has no idea.
Formal verification has always been powerful and almost completely impractical. Theorem is training models to make it 10,000 times faster, and the early results on GPU code and cryptography are hard to ignore.
The pitch is simple: stop guessing who wants to talk to you, and let the machine figure it out.
Every serious SEO tool wants a monthly subscription and your data. Crawler.sh wants neither.
Deep Personality wants to compress years of self-knowledge into a single sitting, and the uncomfortable part is that it might not be entirely wrong.
OpenAI acqui-hired the founder of OpenClaw while Moonshot AI quietly shipped a working version of the product. That's the whole story, really.
AI agents are great at short tasks and terrible at long ones. Aviro built a reinforcement learning layer that teaches agents to stop breaking down after step twelve.
Running GPU servers for AI models is expensive when they're busy and wasteful when they're idle. Cerebrium thinks serverless is the answer, and the economics actually make sense.
The gap between knowing what you want to say and actually writing it down is where productivity apps go to die — Voicr has a specific, almost stubborn opinion about how to close it.
Every deaf or hard-of-hearing person I've talked to about captioning software has the same complaint: it works in exactly one place, and nowhere else.
Building on Solana has always meant fighting your local environment before you write a single line of real code, and Surfpool thinks it has a cleaner fix than anything currently on the table.
The best version of Hacker News has always been the one someone else builds.
The moment before you share your screen is a specific kind of panic, and someone finally built a product for it.
AI tutoring for K-12 math that actually follows the curriculum your kid is already using in class. Not a homework cheat tool. Not a calculator with a chatbot stapled on. A tutor.
Every crypto exchange on earth asks new users to hand over a passport scan, a selfie, and a proof of address — data that then sits on a centralized server, waiting. Zyphe was built specifically to fix that contradiction.
The idea that a single script tag could turn a dead SaaS landing page into a living, breathing room full of avatars is either the best idea in engagement marketing this year or a very confident bet on collective nostalgia.
Referrals already run professional services. The question is whether AI can make the system less dependent on who you happened to grab coffee with last month.
Most AI agents break and stay broken. Giga ML builds agents that learn from their failures in real time and get better without human intervention.
Your AI app can talk to one data source at a time. Airweave wants it to search all of them at once, and the pitch is more practical than it sounds.
Everyone's building AI presentation tools. Anthropic decided to live inside the one you already use.
Finance analysts spend 60% of their time on research drudgery. Rogo built an AI platform that handles the grunt work and somehow convinced 25,000 financial professionals, including teams at major investment banks, to use it.
Most productivity tools try to stop you from wasting time. Shepherd just shows you how much you already have.
Lead generation has always been a numbers game dressed up as a skill. Origami.chat thinks one prompt should do the heavy lifting.
Every AI video tool promises to kill the agency. Reloop's angle is different: it promises to kill the prompt.
Your AI coding sessions vanish into terminal logs nobody can read — one open-source tool thinks that's worth solving.
Voice-to-text has been around long enough to feel boring — Monologue is betting the problem was never transcription, it was everything that came after.
When Arc pulled back, it didn't just leave users without a browser — it left them without a workflow.
Making a software walkthrough used to mean an afternoon of bad takes and worse audio. Guideless is betting there's a better default.
Traditional penetration testing is slow, expensive, and happens maybe once a year. Astra Security built an automated pentest platform that scans continuously and has the compliance certifications to prove the results actually mean something.
Everyone has a notes app graveyard of prompts they'll 'definitely reuse.' Prompt Library is betting $6.35 that there's a cleaner way.
Building an AI agent is the easy part now — deploying it somewhere useful, connecting it to your users, and keeping it running is where things quietly fall apart.
When your codebase is millions of lines spread across hundreds of repos, your IDE's search bar is useless. Sourcegraph built the tool that makes massive codebases navigable, and now they're layering AI on top of it.
Everyone is building AI email tools. Revo is betting that connecting your inbox to everything else you already use is the part everyone else forgot.
Codex-Spark hits 1,000+ tokens per second by running on Cerebras chips — and that partnership might matter more than the model itself.
Everyone can generate a video clip now. Almost nobody can generate ten of them that look like they belong together.
The social media management market is heading toward $164 billion by 2034 — and a Nashville-founded X toolkit just landed #1 on Product Hunt to make its case for a slice of it.
Multilingual meeting AI is a real problem with real incumbents — the question is whether Doraverse has found a wedge or just a feature.
Knowing what software a company runs is only valuable if the data is fresh, sourced, and queryable — which is exactly what most technographics providers have historically failed to deliver.
Process manufacturing plants still run on tribal knowledge and spreadsheet-based recipes. Vesence wants to replace gut instinct with real-time AI optimization, and the founders' path from music production and professional basketball to factory floors is stranger than you'd expect.
A voice AI agent that runs 1:1 onboarding calls sounds like a fever dream from a sales deck — but the underlying problem it's solving is very real.
GPT-5.3-Codex can steer itself through long-running computer tasks, beat the current benchmarks, and — according to OpenAI — debug its own training runs. The question isn't whether that's interesting. It is.
Voice-to-structure apps are a crowded bet, but Sway's refusal to be a transcription tool might be the only interesting thing about the category right now.
Another AI email tool, yes — but the design decision to live inside your existing inbox instead of replacing it might be the only one that matters.
Sameday AI has answered 2 million calls for home services and real estate businesses, booking appointments at a 92% rate. The AI phone agent market is real, and Sameday is one of the companies proving it.
v0 was already useful; now Vercel wants to know if 'useful' can become 'indispensable' before the enterprise window closes.
When your entire pitch is a guarantee, the product better work — and the definition of 'work' matters enormously.
Rather than bet on one model winning, GitHub is building the platform where all of them compete — and that might be the shrewdest move in developer tools right now.
Codex isn't really a coding tool anymore — it's a staffing layer, and that's either exciting or terrifying depending on how many engineers you employ.
Most small businesses run their back office on a patchwork of QuickBooks, spreadsheets, and prayer. Deca wants to replace all of it with a single AI-native ERP platform.
Moltweet is a social network where the users aren't human — and that's either a fascinating research window or a very elaborate screensaver.
A macOS menu bar app that turns dictation into a timesheet sounds almost too obvious — which is exactly why it's interesting.
Every expert, creator, and educator hits the same wall: there's only one of you, and everyone wants your time. Delphi built a platform where you can create an AI version of yourself that thinks, talks, and responds the way you do.
Weather forecasting has been dominated by government agencies running physics simulations on supercomputers for decades. Silurian trained a 1.5 billion parameter model that outperforms NOAA and ECMWF by up to 30%, and they open-sourced the weights.
Freight brokers spend most of their day on the phone with carriers, tracking shipments, and putting out fires. Zeon Systems built an AI that does all three, and it doesn't need a lunch break.
Students are already using ChatGPT to cheat on homework. Excellence Learning wants to give them an AI that actually teaches instead.
Neurological drug development has a 90% failure rate. Exin Therapeutics thinks multimodal AI and high-throughput mouse studies can change those odds.
Grading STEM homework is tedious, time-consuming, and one of the main reasons adjunct professors question their life choices. GradeWiz automates the entire process, handling math notation, code, and handwritten diagrams while providing step-by-step feedback.
Farm labor is expensive, unreliable, and increasingly scarce. Red Barn Robotics built an autonomous weeding robot that does the job humans don't want to do anymore.
Most AI agent tasks are repetitive. Butter records tool call trajectories and replays them deterministically, cutting out the LLM entirely for cases that could have been a script. Faster, cheaper, and more predictable.
Two ex-Revolut builders are turning weeks of regulatory interpretation into minutes. If compliance work feels like reading legislation through a straw, Cardamon is the fix.
Sending money across borders is still slow, expensive, and frustrating. Infinite is building the stablecoin payment rails that make it feel like sending an email.
US immigration paperwork is a black box of compliance risk, attorney bottlenecks, and HR nightmares. Gale is building the platform that automates the worst parts while keeping licensed attorneys in the loop.
Electronic component discovery is stuck in the 1990s. Four ETH Zurich grads are building natural language search across millions of parts, and they have bigger plans from there.
Banks spend months testing credit rules that might not even work. Tejas AI is building a platform that lets risk teams simulate, evaluate, and ship decisioning logic faster than their current process allows.
Most AI tools target knowledge workers who write emails. Macadamia is going after mechanical engineers, construction firms, and manufacturers with AI agents that actually understand CAD files.
Most AI companies are obsessed with text and images. Overstand Labs is building foundational models for audio: speech-to-text, speaker diarization, and classification at scale.
The Pokemon card market has a pricing transparency problem. Misprint is applying financial market structure to fix it, with real-time bid/ask spreads and ML-powered valuations.
Every AI writing tool autocompletes your sentences. Praxim wants to edit your whole document at once, with formatting intact and context from your files and the web.
Investment analysts spend hours digging through SEC filings and earnings transcripts. Fira searches thousands of pages in seconds, cites its sources down to the specific cell, and exports to Excel. The founders previously built products used by 45 million people.
Every doctor needs to be credentialed with insurance companies to get paid. The process is a bureaucratic sinkhole that costs clinics real money when it lapses. Harbera uses AI to keep it all current automatically.
Medical clinics lose patients and revenue to unanswered phone calls every day. Ovlo is an AI answering service that handles scheduling, insurance verification, and prescription refills without making patients sit through hold music.
Affordable housing properties have to verify tenant eligibility constantly or risk losing federal funding. Tire Swing automates the certification and recertification process end-to-end, replacing expensive third-party consultants with AI.
Every AI productivity tool wants to be your copilot. Dex thinks the trick is embedding directly in the browser, learning your context across fourteen apps, and acting before you ask.
No-code tools promised to democratize software. They mostly created a new class of power users who spend their days in Retool. Rebolt thinks the answer is simpler: just tell the AI what you want and let it build the thing.
Salesforce development has a code review problem that generic AI tools can't solve. Sennu AI built a reviewer that understands the full context of a Salesforce codebase.
Moving data between Salesforce, NetSuite, and SAP is the kind of work that makes people quit their jobs. Agentin AI built agents that handle it, and they use reinforcement learning to get better when things go wrong.
FedRAMP compliance is the toll booth between software companies and the $100 billion federal IT market. It takes 16 months and costs over a million dollars. Archon says they can do it in six months and save you seven figures.
Back-office work is the last frontier of automation. CopyCat analyzes how your team actually does manual processes, then builds custom AI agents to handle it without you touching a line of code.
Government procurement is a $700 billion labyrinth that buries good companies in paperwork. Candor's pitch: AI agents that don't just find the opportunities, but actively chase them down.
SAT and ACT prep is a $2 billion industry built on decades-old pedagogy. Miyagi Labs wants to replace the whole stack with an AI tutor that actually adapts to how each student learns.
Law firms spend thousands on advertising to make the phone ring, then lose half those leads because nobody picks up after 5 PM. OpenIntake is an AI system that answers every call, qualifies leads, and converts them to clients around the clock.
Two former pilots are applying computer vision and AI to the problem of keeping planes from hitting each other. The FAA should be paying attention.
Three Stanford engineers are building LLM-powered communications monitoring that promises to catch financial fraud, IP theft, and compliance violations before they become headlines.
Every developer has wasted a week integrating an API that should have taken a day. ReJot wants to replace the SDK-plus-backend-wiring pattern with reusable building blocks that include the data layer. Two ex-Adyen engineers from Amsterdam who spent years thinking about distributed data.
Medical translation in hospitals currently depends on human interpreters who are expensive, slow to arrive, and often unavailable at 3 AM. Vocality Health is building voice AI that lets doctors speak any language in real time, with medical-grade accuracy.
CAD software hasn't had a real interface rethink in twenty years. Adam is betting that text-to-parametric-design can make engineering teams ten times faster. Whether that holds up against SolidWorks muscle memory is the real question.
What if instead of learning where Adobe hid the Gaussian blur, you just said 'blur the background'? Awen is rebuilding creative software around voice commands and AI reasoning. The founder got an O1 visa in visual arts at 21 and ran operations for a luxury fashion company. This isn't a tech bro's idea of what creatives want.
Traditional physics simulations scale exponentially and take thousands of years for certain problems. Trim built a custom transformer architecture that simulates physical systems in constant time. If it works at scale, this changes everything from autonomous vehicles to gravitational wave detection.
A four-person team of PhDs from UCL and Cambridge built an x-ray analysis model that's smaller, cheaper, and more clinically accurate than anything the big labs have shipped. Now they're trying to make radiologists twelve times faster.
GPS, military communications, and intelligence satellites are vulnerable to adversarial threats in orbit. Orbital Operations is building a high-thrust, reusable space vehicle designed to protect them. The founders have resumes from NASA, SpaceX, Blue Origin, and Relativity Space.
Hotels spend a fortune on front-desk staff who answer the same 15 questions all day. Riviera built an AI phone receptionist that handles inquiries, room service orders, and reservations in any language, 24/7.
A two-person team built a DeFi trading platform that's pulling $10M in monthly revenue and $5M in profit. The numbers are absurd, the market is volatile, and the product actually seems well-built.
Character.AI proved millions of people want to talk to fictional characters. Swerve thinks the experience should feel more like a story and less like a chatbot with a costume on.
GPU compute is expensive and orchestrating training jobs across cloud providers is a pain. TensorPool is a CLI that handles the infrastructure so ML engineers can stop being part-time DevOps.
