惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

The Register - Security
The Register - Security
美团技术团队
Recent Announcements
Recent Announcements
MongoDB | Blog
MongoDB | Blog
Jina AI
Jina AI
C
Check Point Blog
aimingoo的专栏
aimingoo的专栏
I
InfoQ
S
Securelist
T
Tor Project blog
GbyAI
GbyAI
L
LINUX DO - 热门话题
V
Visual Studio Blog
AWS News Blog
AWS News Blog
The Cloudflare Blog
腾讯CDC
K
Kaspersky official blog
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
Recorded Future
Recorded Future
李成银的技术随笔
W
WeLiveSecurity
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
M
Microsoft Research Blog - Microsoft Research
G
Google Developers Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
Schneier on Security
Schneier on Security
B
Blog
IT之家
IT之家
爱范儿
爱范儿
H
Help Net Security
Simon Willison's Weblog
Simon Willison's Weblog
NISL@THU
NISL@THU
J
Java Code Geeks
博客园 - 聂微东
T
The Exploit Database - CXSecurity.com
Cyberwarzone
Cyberwarzone
博客园 - 叶小钗
MyScale Blog
MyScale Blog
Application and Cybersecurity Blog
Application and Cybersecurity Blog
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
Project Zero
Project Zero
F
Future of Privacy Forum
D
Darknet – Hacking Tools, Hacker News & Cyber Security
CTFtime.org: upcoming CTF events
CTFtime.org: upcoming CTF events
Hacker News: Ask HN
Hacker News: Ask HN
D
Docker
Apple Machine Learning Research
Apple Machine Learning Research
B
Blog RSS Feed
V
Vulnerabilities – Threatpost

Hacker News: Show HN

Show HN: Physics AI – Visualizing physics problems with step-by-step derivations Show HN: A timeline of recent open source CVE intensity and volume PocketWebTools Show HN: Boatswain, A macOS menu bar app for Fathom Analytics Show HN: Changed boring tip in Claude Code to a joke Show HN: ChangeSpec is an open specification for software change communication Show HN: I built a private, manual 0% balance transfer tracker 3.125-Bit LLM quantization bypassing tensor cores Medical curiosities | Thomas Morris FlutterTime ~ Timezone Planner Steam 上的 Junebug GitHub - Helvesec/rmux: Universal Rust multiplexer with a typed SDK — drive any CLI or TUI app from code. Native on Linux, macOS, and Windows. GitHub - manas15/try-on: LiveLook — Real-time virtual try-on with gesture control, powered by Decart's Lucy VTON model GitHub - vitalysim/the-knowledge-guy: Turn any PDF or EPUB into a structured Claude Code skill - then ask your whole bookshelf a single question. Gemini Omni Flash AI Video Generator | Free Online GitHub - elliotgao2/handsets: A high-performance Android control CLI, built for agents and humans GitHub - enzoferraripapa-arch/ai-vprocess-ops: Engineering memory for AI coding agents: requirements, decisions, evidence, traceability, and V-process/ALM handoff Show HN: Dokkaebi – Run your WASM backend directly on the client side Send messages beyond your lifetime SkinMax App | Your Personal Skin Care Coach GitHub - kmdupr33/fks2g: A CLI for generating LLM-backed metrics for deciding how closely to review code ISS QuietGPT - Make ChatGPT Reply Smaller GitHub - Quintisimo/macfigure: Mac configuration in pkl. Simple alternative to nix-darwin Show HN: SafeRun – Replay debugging and inline prevention for AI agents 3 GitHub - sathvikc/agent-chat-bridge: Turn any AI agent chat session into an async agent. Register a timer, shell command, or webhook — the bridge automatically resumes the session with your prompt when the trigger fires. SnapAPI - Website Screenshot & Data Extraction API Introducing @cipherstash/stack Show HN: E2E Encrypted Terminal Screen Share Windows 98½ Show HN: SafeRun – Replay debugging and inline prevention for AI agents 2 Show HN: My custom Statusline for Claude Code (Python wrapper around claudeline) GitHub - kageroumado/phosphene: A video wallpaper engine for macOS Tahoe Best Remote Jobs — Work From Home | RemoteJobs.place udoc Free AI Rewriter - Revise GitHub - arashThr/hugo-flow: Simple rich-text CMS for Hugo weblogs. Try at https://hugo.arashtaher.com GitHub - light-cloud-com/ice: Free, open-source, visual studio for cloud infrastructure for macOS, Windows & Linux. GitHub - kouhxp/yapsnap: Snap any video URL or audio file into plaintext. No GPU. No cloud. One command. What if we made SIMA2 from Temu iPhone 版“Today” - App Store Runo - Web Scraping API | Any URL to Typed JSON Show HN: AI Editor for Websites GitHub - AdamGonda/ward: Run [ npm i ] safely, audit installs inside a docker container. The Crucible — 8 voices, one verdict Screenshot 2026 05 20 at 4 03 10 PM — Postimages Show HN: Chess Puzzles, but for Developers Show HN: I built Istanbul live transit map Show HN: Agent.email – sign up via curl, claim with a human OTP GitHub - mfairley/expo-callkit-telecom: 📞 CallKit + Core-Telecom for React Native + Expo. A modern react-native-callkeep alternative. I tried 4 LLM speedup techniques on CPU. Three made it slower. Show HN: I made a tool for learning scales, chords, and how to combine them Learn how to build AI products through practice 1 BTC = 17.17 troy oz of gold · Bitcoin Weigh-In p-Hacker — top trending Client Challenge hty GitHub - Artain-AI/ignite-ms: Fast self-hosted embedding engine for search, RAG, and reindexing workloads on NVIDIA GPUs. Built in Rust + TensorRT for teams that care about scale, cost, and control. GitHub - mupt-ai/dari-docs: optimize your documentation through fleets of agents GitHub - dcostenco/prism-coder: The Mind Palace for AI Agents - HIPAA-hardened Cognitive Architecture with on-device LLM (prism-coder:7b), Hebbian learning, ACT-R spreading activation, adversarial evaluation, persistent memory, multi-agent Hivemind and visual dashboard. Zero API keys required. Catio | The Architecture IDE for Modern Software Systems SysWP Radar — Veja TUDO que toca seu site homecrew — package manager for agent skills GitHub - platform-engineering-labs/formae: Infrastructure-as-Code Platform Built for the Future VibeKeys Max - Ready to Ship Show HN: We wrote forensic intelligence reports on 20 open-source codebases GitHub - bytedance/Lance: A 3B-active-parameter native unified multimodal model for image and video understanding, generation, and editing. GitHub - zero-intelligence/zero-protocol: ZERO.md — A universal protocol for personal AI context. Your AI knows your project. It doesn't know you. This file fixes that. Show HN: Chatroom with curl command (requires IPv6) Google Search Console MCP | CalmSEO GitHub - receptron/mulmoclaude: Multi-modal Claude Code Client ArchGenie — AI Cloud Architect for AWS, Azure & Google Cloud GitHub - ueberdosis/hocuspocus: The Yjs CRDT WebSocket backend for conflict-free real-time collaboration in your app. atrium — The shared development environment for humans and agents Show HN: ContextLens- Highlight any text, get context-relevant meaning instantly Show HN: Privacy Focused Voice Journal Free AI Headshot Generator | Professional Photos in 60 Seconds Show HN: Synthra – I got tired of gluing config sources together in Go GitHub - dipankar/dscode: The code editor you can take apart. GitHub - nastronot/sticky-icky: Browser-based sticker design tool for the Zebra LP2844 thermal printer GitHub - rgombash/ocl-nexus-local: Orchestrated Compute Layer for AI agents. Local K3s-based fabric for autonomous code execution, service orchestration, and native MCP tool discovery. LEGO Minifigure Rankings & Ratings One Dev Environment for Humans, Agents and CI Mathabito GitHub - agynio/platform: Agyn is an open-source Kubernetes-native runtime that moves AI agents like Claude Code and Codex from laptops to company infrastructure with the controls enterprises need. GitHub - RaphaelPrevost/hashmap-benchmark Show HN: AI-Powered DNA Research Assistant Show HN: Halftrace – See the shape of your agent's failures from your logs Show HN: Pay-as-you-play game servers from Discord (£2/mo, four games) ProposalKit.io — Proposals clients actually read and sign Show HN: I built a dashboard to combine my WHOOP, Oura and Apple Health data Show HN: I put Codex and Claude in a tank arena; Codex is winning 55% so far Doctor Visit Notes App for Patients and Caregivers | AI Doctor Notes Mashari - Simple AI-first project management Professor Goose - Active Recall AI Study Tutor Show HN: IResearch – C++ search that beat Lucene and Tantivy on their benchmark Document intelligence for AI engineering workflows. Show HN: Widget Cast – Video Widgets for Your iPhone and Apple Watch Raster | Enter the Grid Shitty AI App Idea Generator
What does your investment actually buy? — Post-Money SAFE Calculator
kirushik · 2026-05-21 · via Hacker News: Show HN

A friend-to-friend guide · Post-money SAFE · YC 2018+ form

What does half a million dollars actually buy?

A SAFE — Simple Agreement for Future Equity — turns into shares only when the company raises a priced round. The math is short, the surprises are many. Move the inputs to see exactly how your investment becomes ownership, how it gets diluted at Series A, and what it pays back at exit.

After SAFE converts

5.00%

pre-Series A

After Series A closes

3.65%

your final ownership

Paper value at this round

$0.51M

1.02× at the Series A price

Three snapshots of the cap table

100% = the whole company

Stage 1

Today

Stage 2

SAFE converts

Stage 3

Series A closes

Founders

Option pool

SAFE holder (you)

New Series A investor

i How conversion works

The lower price wins.

When your SAFE converts, you get shares at whichever effective price is lower: the cap price, or the discounted round price. Never both — even though you "have" both terms. Here's how today's inputs play out:

What the "post-money" in post-money SAFE really means

The "post-money" doesn't mean after the Series A money. It means after all SAFEs convert but before the Series A new money. So your $0.5M ÷ $10M cap = 5% is your slice at that moment — and it gets diluted again when Series A comes in. YC's own primer confirms this; it is the most-misread term in startup finance.

ii The dilution journey

From a sheet of paper to a slice of a company.

Five steps. Watch how your percentage moves at each one.

iii What you make at exit

What you walk away with, at any exit price.

Drag the slider — all the way down to a fire-sale if you like. Numbers assume your SAFE has converted into a standard 1× non-participating preferred at Series A, and that all preferred ranks pari passu (equal seniority) — the usual outcome.

$100M

Your payout

$3.65M

3.65% × $100M, after pref check

Return multiple

7.3×

on your $0.5M

IRR (5-yr hold)

48.6%

if exit happens in ~5 years

What "1× non-participating preferred" means for you

At a priced round your SAFE converts into a flavor of preferred stock. On exit, every preferred holder picks the better of two options: take 1× their money back as liquidation preference, or convert and take their share of common. The preference is a floor that only bites when the company sells for less than the stacked-up preferences — at good outcomes everyone just converts. This tool assumes all preferred ranks pari passu; real waterfalls sometimes stack seniority, which changes who gets paid first.

iv What can go wrong

What first-time angels typically miss.

The most expensive mistakes friends-and-family angels make rarely come from the cap or the discount. They come from these:

most important"A cap means I own that fraction."

No: the cap defines your conversion price ceiling, not your final ownership. Your $0.5M ÷ $10M cap = 5% is only your share at one specific moment — right after the SAFE converts and right before Series A money lands. The Series A money and the option pool top-up immediately dilute you further. Headline percentage is a peak; what matters is what's left after the next round.

terminology trap"Post-money means after the round."

In post-money SAFE, "post-money" means after all SAFEs convert — not after the priced round. Y Combinator's own primer states this explicitly. Every first-time investor reads it the other way around.

silent killer"What if there's no Series A — ever?"

SAFEs have no maturity date. If the company chugs along on revenue and never raises a priced round, your SAFE just sits there. No interest accrues, no automatic conversion happens, and you cannot demand your money back. The single biggest under-appreciated risk for friends-and-family investors.

cap-table cascade"More SAFEs just dilute the founder."

In a post-money SAFE, this is true on paper — and that's the problem. Founders who stack four or five SAFEs at different caps wake up at Series A having given away 30–40%. The Series A lead then demands they re-vest, and the SAFE round gets renegotiated. As a SAFE holder you can be on either side of that conversation.

below the cap"What if Series A prices below my cap?"

Then the cap is irrelevant. You convert at the Series A price, the same as the new lead investor (minus your discount, if you have one). The cap is a one-way ceiling, not a floor. It does not protect against down rounds — only against up rounds.

acquisition"What if the company is bought before any round?"

You take the greater of: (a) your money back (the "cash-out amount"), or (b) what you'd get if the SAFE converted at the cap. So a $0.5M SAFE at a $10M cap in a $20M acquisition pays you the conversion path: $0.5M ÷ $10M × $20M = $1.0M. A $5M fire-sale pays you the cash-out: $0.5M back.

no rights"I'm an investor, so I have voting and info rights."

SAFE holders have almost no rights until the SAFE converts. No board seat, no voting, no formal information rights, no pro-rata (unless a side letter says otherwise — and the default post-money SAFE does not include one). For friends-and-family checks under $250K, you should expect none of these.

How this calculator works. Uses Y Combinator's post-money SAFE v1.1 conversion mechanics. SAFE conversion price = min(cap-implied price, discounted round price). Pool top-up is calculated pre-money and dilutes everyone who came in before Series A, including SAFE holders. Exit math assumes SAFE converts into 1× non-participating preferred at the SAFE's conversion price (the standard shadow-preferred treatment). All inputs are kept in $ millions for readability. This tool is for explaining mechanics, not for negotiating. Talk to a startup lawyer before signing anything — most issue mistakes that are visible in a tool like this are caught by them, and the ones that aren't visible are caught by the lawyer.

Things this tool does not model: stacked SAFEs at multiple caps (the most damaging real-world pattern for founders), MFN clauses, pro-rata side letters, QSBS / §1202 tax treatment, anti-dilution provisions in Series A preferred, and the participating-preferred or multiple-pref edge cases that occasionally appear in distressed deals.

Set in Fraunces & Geist. Built, typeset & hosted by Kirill Pimenov, designed hand-in-hand with Claude Opus 4.7 — if the paper‑and‑ink of it looks familiar, now you know whose hand. @kirushik