Startup Videos
The launch videos, explainers, founder stories, pitches, ads, and keynotes founders study before making their own. Browse by what a video is for, then narrow to your industry. Tap any video to play it right here.
-
Aurorin CAD (YC W26) Launch Video
Professional CAD software rebuilt from scratch to be fast and genuinely AI native.
-
AWS and Cerebras team up for fastest possible AI inference
Wafer-scale engines with AWS for record inference speed and scale
-
Cascade (YC W26) Launch Video
Aligns AI models to a company's own proprietary data so agents stay reliable at scale.
-
Castari (YC F25) | Agents Deployment Platform | Launch Video
A deployment platform built specifically for shipping and running AI agents in production.
-
Compresr (YC W26) launch video
An API that compresses LLM context up to 100x without losing the details models need to answer well.
-
InsForge YC Launch Video
Lets AI coding agents deploy and operate an entire backend, database, auth and storage, from a CLI.
-
Introducing Railway App and Platform as a Service
One-click deployment platform for ML models
-
Replicate Review: The Easiest Way to Deploy and Scale ML Models
Cloud platform for running and sharing open-source ML models
-
The Token Company (YC W26) Launch Video
Compression middleware that strips useless tokens from LLM context for faster, cheaper answers.
-
Truly Serverless GPUs, Inside Modal's Fast Cold Starts
Demonstrates truly serverless GPUs cold starting over 40x faster via a custom container filesystem and CUDA checkpoint restore.
-
What is PostHog? (Official Demo & Tutorial)
Official walkthrough of the open source product analytics platform: replay, flags and A/B tests in one.
-
bolt.new launch
Prompt to a running full stack app in one unbroken take, the demo that pushed vibe coding into the mainstream.
-
Claude x Blender MCP launch
A silent, sped up 23 seconds of Claude driving Blender lets the aha land without a single word.
-
Convex is changing backend
A reactive backend where queries push live updates to clients with no manual websocket or cache wiring.
-
ElevenLabs launch
Stacking uncannily human voices across languages back to back makes you double check it is not a real actor.
-
Embedder Launch (YC S25)
Brings AI coding agents to embedded systems and firmware, a corner most coding assistants ignore.
-
Fine-tune AI video models on Replicate
Launch demo for fine tuning open video models, turning any YouTube URL into a dataset and training in minutes via one API run.
-
Gecko Security: Product Hunt Launch
An AI security engineer that finds exploitable bugs, it uncovered 30 plus CVEs in Ollama, Gradio and RAGFlow.
-
Groq's LPU Re-Architects AI Inference for Low-Latency Performance
Architectural innovation reducing latency from seconds to milliseconds
-
Inside Together AI: The Research Driving Next-Gen Open-Source Models
Platform for building and running open-source generative AI models
-
Introducing Agentic Editing in Zed
Agentic editing inside the fast Rust editor: the agent reads, writes, and runs code in your project.
-
Introducing Claude Code
Terminal demo: Claude edits files, runs tests, and opens a PR across a real repo from one prompt.
-
Introducing GPT-5
The team demos GPT-5 uniting reasoning and speed with live coding and health answers; how a lab frames a flagship consolidation launch.
-
introducing lipsync-2, the most natural lipsync model in the world
Zero shot lipsync that edits anyone's mouth to match new audio while keeping their style, no per person training.
-
Introducing Metal: NVMe Drives with Unlimited IOPS
Locally attached NVMe nodes giving unlimited IOPS and far lower p99 than network storage.
-
Introducing Seer: Sentry's AI Debugging Agent
An AI agent that root causes errors from real issue data and proposes code fixes.
-
Launch HN: Manufact (YC S25) - The MCP cloud to get your product on Claude and ChatGPT
Cloud infrastructure for shipping MCP servers straight into the ChatGPT and Claude app stores.
-
Lemma YC F25 - Launch
Catches AI agents degrading in production and automatically opens a pull request with the fix.
-
Lovable launch
Prompt to a shipped app inside the clip, the sheer speed is the entire argument for the product.
-
MangoDesk (YC S25) | Launch Video
Generates bespoke AI annotation pipelines in seconds so teams can build custom evals and training data fast.
-
Massdriver cloud infra launch | Massdriver
Diagram based demo provisioning cloud infra from pre approved modules instead of raw Terraform, with guardrails enforced.
-
Mesmer YC Launch Video
Watches code, specs and meetings to hand CTOs a live weekly picture of what actually shipped.
-
Observee (YC S25) | Launch Video
Lets builders wire AI agents into 1,000+ tools with managed auth, security and observability in one SDK.
-
Onyx: The open-source chat UI (Demo)
Open source ChatGPT style workspace that works with any LLM and plugs into your company docs, apps and search.
-
OpenAI Codex in ChatGPT in 5 Minutes
Delegate parallel coding tasks to a cloud agent that runs each in its own sandbox and files PRs.
-
OpenAI Devs launch
A crisp dev keynote cut where the new API ships a working agent on stage, capability over talk.
-
OpenSesame launch
A drop from space intro and a sharp thesis to camera give a dev tool the polish of a hardware keynote.
-
Replit Agent launch
Typing an app idea and watching it deploy live turns anyone can build from slogan into on screen proof.
-
Stytch Demo: Auth for Users and AI
Modern identity infrastructure supporting both users and AI agents
-
The Birth of Contextual AI - RAG 2.0
A recap of Contextual AI's RAG 2.0 launch, built for production grade accuracy over unstructured data
-
The Future of Development: Introducing Warp 2.0
Reframes the terminal around AI agents, showing multi step coding tasks run and reviewed inside the command line.
-
Together AI on the Shift to Smaller, Purpose Built Models
Infrastructure platform for specialized AI models optimized for tasks
-
Twilio Brings Customer Engagement Into the AI Era
Communications platform enables AI-first customer engagement at scale
-
Warp 2.0: The Agentic Development Environment
Turns the terminal into an agent surface: prompt a task and watch it run commands and fix its errors.
-
Webild launch
An app assembling itself live from a spoken idea keeps the vibe coding promise literal and fast.
-
YC Launch Video: VibeFlow
Generates full stack apps from a prompt with an editable, visual backend instead of a black box.
-
Your First Personal Robot
Launches K-Bot, a full size open source humanoid built in a garage and priced at 8,999 dollars for developers.
-
Announcing Deno 2
Deno 2 keynote: full Node and npm compatibility plus deno compile to a single executable.
-
Cartesia Sonic launch
A state space text to speech model built for roughly 135ms latency, with multilingual voices.
-
Claude, Computer use for coding
Claude moves the cursor, clicks, launches a server, and fixes its own bug building a site; the first mainstream agent driving a desktop.
-
Code-Like Database Branching for Postgres with Neon
Git style instant Postgres branching: copy on write branches of a live database created in seconds.
-
Cognition AI just introduced Devin, the most capable autonomous coding agent
First AI software engineer autonomously solving real tasks
-
Cursor 3 minute demo
A tight three minute screen demo of an AI code editor predicting edits and answering the codebase, no hype.
-
Gemini 2.0 Flash, our new AI model for the agentic era
Native image and audio output plus live tool use in one model; Google's opening move into agentic, multimodal generation.
-
Greptile: AI That Actually Understands Codebases
An AI code review agent with full codebase context, used by PostHog and Raycast to catch bugs and merge faster.
-
Groq runs Mixtral 8x7B at 500 tokens per second (LPU demo)
A full LLM answer appears faster than you can read it on the LPU; the speed demo that reframed inference latency as a headline feature.
-
Groq-LPU Inference Engine Better Than OpenAI ChatGPT And NVIDIA
Custom LPU hardware achieving record low-latency LLM inference
-
Hugging Face - The AI community building the future
Community hub and platform for open-source AI models and datasets
-
Introducing Devin
Showing Devin fix a real Upwork job end to end unassisted made the first AI software engineer claim feel earned.
-
Introducing Reka multimodal AI coding and code execution capabilities
Reka launches coding and code execution abilities for its multimodal models, shown live
-
Launch YC: Tusk (YC W24) UI/UX Fix Demo
An AI coding agent that picks up annoying UI and UX tickets and opens the pull request while engineers focus elsewhere.
-
Meet the Empathic Voice Interface (EVI)
Pitched as the first voice AI that reads the emotion in your voice and responds to it.
-
Meta announces Llama 3
Meta unveils open weight Llama 3 to a developer room, the release that kept the strongest open model in the race against closed labs.
-
Palmier (YC S24) Launch Video
A video editor rebuilt for AI, the same tool used to cut cinematic launch videos for dozens of other YC startups.
-
PhonePe launches Indus Appstore, challenger to Google Play
PhonePe unveils the made in India Indus Appstore, a direct challenger to Google Play, at its Delhi-NCR launch.
-
Postbot | Postman AI assistant
Postman introduces an AI assistant that writes API tests, docs and scripts from a plain prompt.
-
Software supply chain security with Socket
Prevent dangerous packages from entering your software supply chain
-
Supermemory launch
A developer piping scattered notes into one recall API, the demo doubles as the docs devs trust.
-
The next generation of Linear Projects
Redesigned project planning from earliest stage through release
-
The Windsurf Editor (World's First Agentic IDE)
First agentic IDE with autonomous code generation and multi-file awareness
-
A Sneak Peek at Copilot X
The 2023 vision reveal: chat in the editor, PR descriptions, and docs Q and A, past autocomplete.
-
Aider: AI pair programming in your terminal
Terminal pair programmer that edits files and auto commits each change straight into your git repo.
-
Auto-GPT demo (autonomous GPT-4)
The viral April 2023 demo: GPT-4 loops on its own goals, writes code, and self corrects unattended.
-
Bun 1.0 is here
One runtime that runs, bundles, tests, and installs packages, drop in compatible with Node.
-
GPT-4 Developer Livestream
Brockman live codes with GPT-4 and turns a hand drawn napkin sketch into a working website, proving vision plus reasoning in one take.
-
Introducing Cody, Sourcegraph's AI coding assistant
Cody answers whole codebase questions using the code graph, not just local files.
-
Introducing Palantir AIP, Capabilities and Product Demo
Live demo of putting LLMs to work securely on real operational data inside Palantir's ontology.
-
Introducing v0: Generative UI
Type a prompt, watch React plus shadcn UI code stream out live: the demo that made generative UI feel real for developers.
-
Lago open source billing launch | Lago
Screen demo configuring usage based pricing in Lago's open source UI, going from metered events to a finished invoice.
-
Multion Demo
An early demo of MultiOn's browser agent completing a multi step task end to end on its own
-
Raycast 101: The launcher
Sells a keyboard launcher on raw speed: every action one hotkey away, no mouse, replacing Spotlight for power users.
-
What the Fly? What is Fly.io, anyway?
Explains running full stack apps as fast booting micro VMs on hardware placed close to users.
-
ACT-1 (Adept AI) Demo (2022.09.14)
The original ACT-1 demo of Adept's action transformer clicking and typing through real web apps
-
Hello Pinecone: Launching Your First Vector Database
The hello world for the managed vector DB: create an index, upsert embeddings, run a similarity query.
-
Requestly network debugging launch | Requestly
Shows intercepting and rewriting a live Android app's network requests with no proxy setup, a debugging trick devs love.
-
Arengu signup flow builder launch | Arengu
Product intro building a multi step signup flow with validation and API calls visually, no backend code, then embedding it.
-
Axolo GitHub Slack code reviews launch | Axolo
Demo of a GitHub pull request spinning up a dedicated Slack channel so reviewers discuss and merge without context switching.
-
hotglue user-facing integrations launch | hotglue
Founder wires up a Salesforce and GitHub sync in minutes using hotglue recipes, no custom ETL code written by the developer.
-
Lightly ML data curation launch | Lightly
June demo where self supervised selection picks the most valuable one percent of a dataset to label, cutting annotation cost.
-
Sieve video AI platform launch | Sieve
December 2021 walkthrough where the founder runs a search query across raw video and pulls matching clips from the dashboard.
-
Webiny serverless CMS launch | Webiny
Overview walking through the serverless framework plus the drop in headless CMS, all deploying onto AWS Lambda in one project.
Know a video every founder should study? Suggest a video
Questions, answered
What startup videos should founders watch?
Founders on the eChai network recommend videos across launches, explainers, founder stories, pitches, ads, and keynotes. The library ranks them by how many founders recommend each.
Where can I find great startup launch and explainer videos?
The eChai /videos library is a curated, founder-recommended shelf of the best startup videos, browseable by purpose, industry, and region.
