LAUNCH·INDEX
LNCH1,007+7.6%

Experimental: traction = early-adopter votes (PH) / points (HN), not downloads or usage; updated about once a day and frozen ~31–36h after launch on Product Hunt (7 days on Show HN); sectors come from keyword rules (~70–75% accurate); data starts Oct 3 (HN) / Oct 9 (PH). “—” means not enough data; nothing is estimated. How it works

Stocks

Every launch is a stock: its traction is its votes, its sector is its category, and LNCH is the whole market. How it works

Data as of Oct 17, 2026, 12:21 PM UTC · traction last updated Oct 17, 13:13 UTC · updates about once a day

Get the weekly Launch Index

Every Tuesday: which launches broke through, which categories are heating up. Measured, not sponsored.

Free, every Tuesday. Unsubscribe anytime.

LNCHLaunch Index · all sectors
3,716+576 (+18.3%)

launches · Oct 14–Oct 16 vs Oct 11–Oct 13

Hit rate (settled)
6.3%
— · Oct 12–Oct 14 vs Oct 9–Oct 11
Trend · Volatility
+8.4 · 33%
228 hits of 3,645 settled
daily launches, 8 complete days

Launches: Weekday effect likely: Wed–Fri is compared with Sun–Tue, not the same weekdays last week (not enough history yet; expected from about Oct 19). Launch counts and hit rates swing a lot by weekday.Hit rate: Weekday effect likely: Mon–Wed is compared with Fri–Sun, not the same weekdays last week (not enough history yet; expected from about Oct 21). Launch counts and hit rates swing a lot by weekday.

Launched on Oct 16 (the latest complete UTC day) or listed since the previous run (IPO), ranked by current traction. Product Hunt launches all list at 07:01 UTC; their traction has had at most one or two daily updates. 127 matches. Spam giveaways excluded.

Mixed sources. Traction is Product Hunt votes (PH) or Show HN points (HN) — not the same unit, and PH has ~7× the launches. So each source is ranked on its own and the two are alternated (PH #1, HN #1, PH #2, HN #2, …); the # column is that combined order. Pick a source above to see one source’s full ranking.

#SectorSrc
1Harness by Autonomous

Every coding agent, every machine, one command center.

AGNTPH326+26▲ +8.7%135h ago
2rStation

A shared research lab for your AI agents

AGNTHN53+3▲ +6.0%027h ago
3NitroStack

Build, test, and deploy MCP servers in one place

AGNTPH159+159IPO—1311h ago
4Weatherdeck

a read-only status board for AI coding agent lanes

AGNTHN9+1▲ +12.5%0023h ago
5Dottr

Agent that drafts contracts, flags risks + gets them signed

AGNTPH120+9▲ +8.1%535h ago
6Play a 5 minutes match of CS 1.6 1v1 while you wait for your agentAGNTHN500.0%020h ago
7Parallax

Every coding agent you run, in one place.

AGNTPH99+1▲ +1.0%135h ago
8PTrace

An iPhone Running Tracker with MCP

AGNTHN30—030h ago
9Notion: Context for your agents

Switch AI agents without re-explaining your work

AGNTPH94+94IPO—111h ago
10Deterministic AI Agents GuardrailAGNTHN30—0030h ago
11BOSS Console

A tiny OS for your always-on agents

AGNTPH93+1▲ +1.1%035h ago
12Heldkeys, see which credentials your AI coding agent can reachAGNTHN20—028h ago
13Ultracite v8

The linter and formatter for agentic development

AGNTPH82+82IPO—211h ago
14Busabase

an open-source system of record for AI agent work

AGNTHN20—0027h ago
15Craft Agents

The world's most powerful agents, with the UX they deserve

AGNTPH78+78IPO—311h ago
16We built an open source, agent-first, alternative to ProfoundAGNTHN20—0026h ago
17Nemotron 3.5 Super VL

NVIDIA’s open multimodal model for specialized agents

AGNTPH77+5▲ +6.9%035h ago
18Termexo

one AI operating your other coding agents, on Windows

AGNTHN20—0026h ago
19Step 5 Preview

600B open weights, trained for long-running agents

AGNTPH71+71IPO—611h ago
20Beeline Multiplayer Agent HarnessAGNTHN20—0039h ago
21Tetriz Harness

Turn what your team learns into everyone’s agent setup

AGNTPH67+5▲ +8.1%135h ago
22I built 7 B2B APIs and 5 MCP servers for compliance and tradeAGNTHN20—0035h ago
23Agent Vault

Superpowers for your Agents

AGNTPH6300.0%035h ago
24A music video made end to end by Claude Code, full production repoAGNTHN20—0032h ago
25Neophyte

Watch an AI agent sign up and onboard to any product

AGNTPH5900.0%035h ago
26AegisKernel

Zero-trust deterministic security runtime for AI agents

AGNTHN20—0029h ago
27Scalebrowser

Human browser capabilities. Now for AI agents.

AGNTPH57−1▼ −1.7%035h ago
28Claude Code, deploy using thevibehosting.com == Your app on root LinuxAGNTHN1+1IPO—29h ago
29CrewTower Mod for Claude Code

Talk to Claude Code right from your MacBook notch

AGNTPH57+57IPO—311h ago
30Autonomous community research lab powered by agentsAGNTHN10—026h ago
31Toad

Point your agent at any machine you own.

AGNTPH5500.0%035h ago
32AI agent follows instructions you never approvedAGNTHN1+1IPO—0014h ago
33Atelier

The studio your AI coding agents work in

AGNTPH5500.0%035h ago
34Agentfork

Replay an LLM agent from any step and test fixes

AGNTHN1+1IPO—0010h ago
35Palomino

Ultrafast workspace for agents to run in parallel

AGNTPH5500.0%035h ago
36Offscreen

Give your agents a Mac screen of their own

AGNTHN1+1IPO—006h ago
37AI Checker by Signum.AI

Check if AI agents can see your website for free

AGNTPH9+1▲ +12.5%035h ago
38RAE FAANG

catalog and CLI for building products with coding agents

AGNTHN1+1IPO—006h ago
39Nemotron 3.5 Bolt

Run everyday agentic tasks locally on a 24GB GPU

AGNTPH9+9IPO—111h ago
40Gent, email a task to Codex or Claude CodeAGNTHN10—0027h ago
41Probe

Native API client for humans and agents

AGNTPH500.0%035h ago
42Agent Chat in Your Markdown Notes

Chat with Context and Save Response

AGNTHN10—0026h ago
43 NVIDIA Nemotron 3.5 Super VL

It Powers Specialized Agents with Multimodal Reasoning

AGNTPH40—035h ago
44PocketConsole

Run Claude Code on your own server from your iPhone

AGNTHN10—0026h ago
45Readdit Later

The AI agent that helps you learn from your Reddit saves

AGNTPH40—035h ago
46I benchmarked 6 MCP proxies in Claude Code, then built my ownAGNTHN10—0021h ago
47Cafecito

A coffee shop your AI agent can order from

AGNTPH30—035h ago
48CtxWise

MCP schema tree-shaker and firewall for Claude and Codex

AGNTHN10—0019h ago
49Warden

Lock your Mac’s input. Let your AI agents keep working.

AGNTPH3+3IPO—111h ago
50Control how much your AI agent can spend with KeydrisAGNTHN10—0032h ago

Traction = current votes (PH) / points (HN) — early-adopter attention, not downloads or usage. Gained = traction gained between the two latest daily ingest runs; Gain % only when the previous traction was 5+. Buzz = comments gained in that interval. “frozen” = no longer tracked, so no current gain. How it works

Launch Index