Tony Wang22 min readProduct Hunt Trends 2013–2026: How AI Agents Took Over Startup Launches
Product Hunt trends, 2013–2026: we read every leaderboard. AI agents went from 0 to 9 of the annual Top 10 by 2025, and the trend rotates monthly, not yearly.
Most "state of tech" takes are vibes. Product Hunt's leaderboard is one of the few public, longitudinal records of what people actually shipped and what other builders actually upvoted — every day since 2013. So instead of guessing, we read the boards: every annual leaderboard from 2013 to 2026, plus a stack of monthly ones, and tallied what each year was about.
Ryan Hoover launched Product Hunt on November 6, 2013 as a Thanksgiving-break email list, so the board is a ~12.5-year record — and that's the more dramatic span, because the entire "AI everything" earthquake is compressed into its back third.
A method note, with some irony: we pulled the leaderboards programmatically through a structured web-data API. Trying to read the same pages in a plain browser hits a Cloudflare "verifying you are not a bot" wall — which is exactly why reliable, structured access beats hand-scraping HTML. (Caveats are at the end; the short version: the annual board is a curated top-~20, scores are cumulative upvotes, and 2026 is year-to-date.)
Here's the whole twelve years in one shot — Product Hunt's five leading product categories, racing for the top each year. Watch Artificial Intelligence climb from a sliver to first place, overtaking Design Tools and Productivity:
Show the data
| Topic | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AI | 4.5 | 6.6 | 3 | 5.8 | 4.1 | 3.2 | 7.2 | 6.8 | 11.7 | 35.9 | 42.1 | 49.8 | 46.8 |
| Productivity | 31.8 | 23 | 27.7 | 31.7 | 36.3 | 40.6 | 40.2 | 44.9 | 46.7 | 32 | 25.1 | 23.2 | 28.2 |
| Design Tools | 45.5 | 44.3 | 37.6 | 37.5 | 32.7 | 26.7 | 25.8 | 22.7 | 14.7 | 9.1 | 11.3 | 8.7 | 4.8 |
| Developer Tools | 18.2 | 18 | 20.8 | 15 | 15.8 | 16.6 | 11.3 | 11.6 | 15.2 | 10 | 13.8 | 12.2 | 13.7 |
| Marketing | 0 | 8.2 | 10.9 | 10 | 11.1 | 12.8 | 15.5 | 14 | 11.7 | 13 | 7.7 | 6.1 | 6.5 |
From mobile apps to an AI monoculture: the 12-year arc
The board's center of gravity moved in clear eras:
- 2013–2014 · Mobile app discovery. Almost everything is tagged generically "Tech," with a heavy iOS lean. 2013's top product is Sunrise (a calendar app later bought by Microsoft); 2014's #1 is literally Product Hunt's own iOS app. The platform is a place to find apps.
- 2015–2016 · The maker era. Curated directories and founder toolkits dominate — 2015's #1 is Startup Stash (400 startup tools), 2016's is Startup Pitch Decks. AI first pokes through as novelty (Prisma, Amazon Go).
- 2017–2019 · Design tools & productivity. The two most durable topics in the whole dataset. AutoDraw (2017), Notion 2.0 and remove.bg (2018), Checklist Design (2019). AI is still toys (AutoDraw, the "Not Hotdog" meme app, Google Duplex).
- 2020 · The remote-work pivot. HEY tops a board full of video-chat and remote tooling; the community's Golden Kitty Product of the Year is Clubhouse.
- 2021 · The no-code peak — the calm before the storm. 2021's #1 is Cal.com, surrounded by no-code site builders. AI products in the Top 10: zero.
- 2022–2023 · The AI spark, then the breakout. ChatGPT debuts on the 2022 board with the single highest score; by 2023 "Artificial Intelligence" is the #1 topic and the Golden Kitty goes to ChatGPT itself.
- 2024–2026 · The agent era. 2024's #1 is Wordware — a tool for building AI agents. By 2025, 9 of the Top 10 are AI; in 2026 the incumbents arrive (Anthropic's Cowork, ChatGPT Health) and the unit of competition becomes the agent, not the app.
| Year | Most-upvoted #1 | Top topics |
|---|---|---|
| 2013* | Sunrise (calendar) | Tech, iOS |
| 2014 | Product Hunt for iOS | Tech, iOS, Web App |
| 2015 | Startup Stash | Web App, Design Tools |
| 2016 | Startup Pitch Decks | Web App, iOS, Tech |
| 2017 | AutoDraw (Google) | Tech, Design Tools, Dev Tools |
| 2018 | Notion 2.0 | Design Tools, Productivity |
| 2019 | Checklist Design | Design Tools, UX, Productivity |
| 2020 | HEY (email) | Productivity, Design Tools |
| 2021 | Cal.com | Productivity, Design Tools |
| 2022 | ChatGPT (top score) | Productivity, AI, SaaS |
| 2023 | Chat by Copy.ai | AI, Productivity, Web3 |
| 2024 | Wordware | AI, Productivity, Dev Tools |
| 2025 | Screen Studio 3.0 | AI, Productivity, Design |
| 2026† | Cowork (Anthropic) | AI, Productivity, Dev Tools |
* 2013 is a partial first ~2 months. † 2026 is year-to-date.
AI didn't trend — it took over
Watch the whole board's category mix reshape itself. Each band below is one of Product Hunt's five leading product categories (deep sample, ~180–200 products/year); Design Tools and Productivity own the early years, then the orange Artificial Intelligence band — barely a sliver before 2022 — swells into the widest by 2024:
Show the data
| Topic | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Artificial Intelligence | 4.5 | 6.6 | 3 | 5.8 | 4.1 | 3.2 | 7.2 | 6.8 | 11.7 | 35.9 | 42.1 | 49.8 | 46.8 |
| Productivity | 31.8 | 23 | 27.7 | 31.7 | 36.3 | 40.6 | 40.2 | 44.9 | 46.7 | 32 | 25.1 | 23.2 | 28.2 |
| Design Tools | 45.5 | 44.3 | 37.6 | 37.5 | 32.7 | 26.7 | 25.8 | 22.7 | 14.7 | 9.1 | 11.3 | 8.7 | 4.8 |
| Developer Tools | 18.2 | 18 | 20.8 | 15 | 15.8 | 16.6 | 11.3 | 11.6 | 15.2 | 10 | 13.8 | 12.2 | 13.7 |
| Marketing | 0 | 8.2 | 10.9 | 10 | 11.1 | 12.8 | 15.5 | 14 | 11.7 | 13 | 7.7 | 6.1 | 6.5 |
Plot AI against Productivity — the most durable old-guard category — as a share of each year's ~190 top products, and you can watch the handover happen. Productivity is the steady incumbent (~30–47%); AI goes from a rounding error to dominant, crossing it in 2023:
The spark was the back half of 2022, when ChatGPT launched and immediately posted the top score on the board. From there AI's share of the board roughly quadrupled in a single year — ~12% in 2022 to ~42% in 2023 — then climbed to ~53% (2024) and ~70% (2025). Just as telling is how the flavor of AI kept mutating: chat/LLM wrappers (2023) → AI agents and vertical apps (2024) → agentic workflows, AI dev-teams and "vibe coding" (2025) → incumbent agents and MCP-native tooling (2026). The category didn't just grow; it kept reinventing what "an AI product" even is.
A single month, four eras
Zoom from years to months and a paradox appears. At the product level, the monthly Top set turns over almost completely — across the six monthly boards of 2026 so far, essentially no individual product repeats in the Top 8. But at the theme level, almost nothing changes: every 2026 month is ~100% AI/agent products. What recurs is brand families, not entries — Anthropic's Claude charts nearly every month via a different launch; the "…Claw" agent family propagates across the year.
Take one month per era and the trajectory is stark:
browsers, window managers, photo libraries
COVID: meditation, remote 1:1s, Zoom tools
GPT-3-era 'ChatGPT for X' wrappers
autonomous agents that do a job
So the thing that "changes fast" in 2026 isn't the category — it's the sub-flavor of AI, month to month: January was agent runtimes and code review; March, AI design and agent marketplaces; April, meeting-notes and evals; May, agent-commerce and voice-calling agents; June, fundraising and GTM agents. The trend clock compressed from a theme per year to a sub-theme per month.
The platform grew, then the clock sped up
Product Hunt itself scaled fast and then plateaued — which makes the acceleration of trends (not volume) the real story:
What actually changed in 12 years
- From tools you use to agents that do it for you. 2013–2021 winners are utilities a human operates; 2024–2026 winners are described as employees — "the AI to-do list that does itself", "a team of AI agents that runs your stores". The verb moved from use to delegate.
- Topic diversity collapsed into a monoculture. Through 2021 the top topics were a genuine mix (Design, Productivity, Dev Tools, No-Code, Hardware, Crypto). Since 2024, "AI" is on nearly everything; the other topics are just the domain the AI points at.
- Durable winners gave way to franchise churn. Early hits stayed relevant for years; the 2026 board is version bumps and brand families cycling through the top spots monthly.
- Each wave left a fingerprint, then receded — except one. Hardware, the maker boom, crypto/Web3, COVID/remote, no-code — each surged and faded. AI is the first wave that, instead of receding, became the substrate the next waves run on.
The takeaway for anyone shipping: the leaderboard rewards whatever frame is ascendant. In 2019 that was a polished design resource; in 2026 it's an autonomous agent with a clear "it does the job for you" promise. The same capability launched in the year's language outperforms one launched in last year's.
How the board itself changed — not just the topics
The topics flipped to AI, but three quieter shifts in the leaderboard's own mechanics are just as telling.
Winners now keep gaining votes after launch day. Comparing each year's winners' current upvotes to their launch-day score: early-era winners were launch-day spikes that bled votes afterward (ratio below 1); since 2025 they keep climbing (above 1) — the board rewards sustained momentum over a one-day burst.
Annual winners cluster in Q1. Across 2018–2025, a disproportionate share of year-topping products launched early in the year — February is the single peak and Q1 holds ~37% of winners. Early launches have the most time to accumulate the cumulative votes that decide annual rank.
AI didn't just grow — it fused with Productivity. Among each year's top winners, the number carrying both the AI and Productivity tags went from ~zero to 9 of 17 in 2025 — "AI + Productivity" became the board's defining combination, and the old Web3/blockchain pairings vanished after 2023.
Where are Product Hunt's winners now?
A leaderboard tells you what launched, not what lasted. So we ran the winners back through the same API: of the products that topped Product Hunt, how many are still alive today? We fetched the homepage of 28 era-defining winners and pulled each one's current traffic.
The catch — and it's the whole point — is that a domain returning a page is not a living product. Four "winners" still answer with 200 OK but serve something else entirely: Station's domain is now a Thai casino, Sunrise's is an SEO content farm, Rewind's is a generic AI-tools page, and Polywork's is a different site builder. Traffic estimators even keep serving the old title, so they'd happily score these as alive. Only fetching the live page and cross-checking the traffic — the same fetch-and-verify discipline behind our anti-bot index — reveals that the product is gone.
By that bar, here's how the 28 winners split:
Roughly seven in ten still run the product that won — but about three in ten are dead or have pivoted away from the thing that put them on the board. A #1 badge was no insurance: Sunrise (2013's beloved calendar, bought by Microsoft and shut down) and Wordware (2024's literal #1 — "$30M, the largest seed in YC history" — now replaced by a different product) both topped Product Hunt and are effectively gone.
And the deaths aren't evenly spread — they cluster hard in one era:
Sunrise, Prisma
Station
pandemic / social / novelty bets
too recent to judge
Put both dimensions together — which era each winner launched in, and where it ended up — and the whole story sits in one flow:
Show the flows
| 2017–19 → Still running | 6 (21.4%) |
| 2013–16 → Still running | 4 (14.3%) |
| 2020–22 → Gone | 4 (14.3%) |
| 2023–26 → Still running | 4 (14.3%) |
| 2020–22 → Still running | 3 (10.7%) |
| 2013–16 → Gone | 2 (7.1%) |
| 2013–16 → Diminished | 1 (3.6%) |
| 2017–19 → Gone | 1 (3.6%) |
| 2020–22 → Diminished | 1 (3.6%) |
| 2023–26 → Diminished | 1 (3.6%) |
| 2023–26 → Gone | 1 (3.6%) |
Half of the 2020–22 cohort is already gone — the pandemic-and-social-novelty winners (mmhmm, Polywork, Typedream, Rewind, plus a much-diminished Clubhouse). The survivors, in every era, are the trend-independent ones: single-purpose utilities (Workflowy, Coolors, Hunter, remove.bg) and infrastructure (Notion, Supabase, Cal.com, Resend). Winning a launch is a moment; outliving the trend that powered it is the hard part.
And outliving the trend still isn't the same as winning the market. Among the survivors, monthly traffic is brutally top-heavy:
Winning a Product Hunt launch gets you on the board; becoming Notion is a different game entirely.
How to use this before you launch
The leaderboard rewards trend-fit, but the data above says trend-fit and durability are different games. If you're timing a launch, treat these as the operating rules:
- Don't launch as generic "AI" — launch as this month's sub-flavor. "An AI tool" is invisible on a board that's already ~100% AI. The launches that break out name the specific job that's cresting right now — code review, meeting notes, agent commerce, GTM — so match the sub-theme that's ascendant the month you ship.
- Phrase the promise as delegation, not tooling. The winning verb moved from use to delegate: "does X for you," not "a tool for X." "The AI to-do list that does itself" beats "a smarter to-do list."
- Use trend-fit for launch velocity, but ship a utility for survival. Trend-of-the-moment framing wins the day; the products still alive years later are single-purpose utilities and infrastructure (Notion, Supabase, Cal.com, remove.bg). Ride the wave to get seen — but build something people still need after it breaks.
- Time it: track the monthly sub-theme rotation, and aim for Q1. The AI sub-flavor shifts roughly monthly, so launch into the one that's cresting. And if you're chasing the year-end lists, launch early — February is the single peak month for annual winners, because early launches have the longest to accumulate the cumulative votes that decide annual rank.
- Treat Product Hunt as an awareness and positioning signal — not proof of retention or market leadership. A #1 badge is a moment of attention, not evidence of a durable business; among survivors, traffic is brutally top-heavy. Winning the launch and becoming Notion are different games.
How we did this (and the caveats)
We read Product Hunt's annual and monthly leaderboards directly via a structured API, then cross-checked the platform facts — launch date, launch volumes, and Golden Kitty winners — against public sources. Caveats worth stating plainly: the annual board is a curated top-~20, not an exhaustive vote ranking; scores are cumulative upvotes, so they're directional, not exact annual totals; 2026 is year-to-date; 2013 is a partial ~2 months; and "AI in the Top 10" counts are by listed topic, so borderline cases are judgment calls — but the direction is robust regardless. For the topic charts (the race, streamgraph, and AI-vs-Productivity line) we went deeper than the annual top-20: we aggregated all twelve monthly leaderboards per year and de-duplicated by product, giving a ~180–200-product sample each year (vs ~17 at the very top) — which is why the AI share here (e.g. ~70% in 2025) is a touch lower and far smoother than a top-of-board figure would be. The yearly endpoint's deep pagination is broken, so the survival, vote-accumulation, seasonality, and co-occurrence cuts use the top-~17/year set. For the "where are they now" check, we fetched each winner's homepage and pulled its current SimilarWeb traffic, counting a product as gone only when the live page no longer serves it — a domain that returns 200 with unrelated content (a parked page, a casino, a different product) is dead, not alive. That liveness snapshot is June 2026, and "gone" includes products that pivoted away from the winning idea, not only outright shutdowns.
The datasets behind the piece, at a glance:
| Dataset | Sample size | Used for | Caveats |
|---|---|---|---|
| Deep monthly-aggregated sample | ~180–200 products/year (all 12 monthly boards, de-duplicated) | The topic-category charts: race, streamgraph, AI-vs-Productivity line | Smoother and a touch lower than a top-of-board figure (AI is ~70% in 2025, not 90%); "AI" tagging is a judgment call on borderline products |
| Annual top boards | ~17 products/year | Survival, vote-accumulation, Q1 seasonality, AI+Productivity co-occurrence | Curated top list, not a full vote ranking; cumulative-upvote scores; the deep-pagination endpoint is broken, so these cuts cap at the top ~17 |
| Monthly boards (era sample) | Top 8 per sampled month | Sub-theme rotation and product-level churn | Sampled months, not every month; 2026 is year-to-date |
| Winner survival + traffic | 28 era-defining winners | Liveness, pivots, and surviving-product traffic | Live homepage fetch + SimilarWeb (May–June 2026); a 200 with unrelated content counts as gone; "gone" includes pivots |
If you're wondering how we read pages that block bots: that's the day job. Crawlora is a web-data API for AI agents and pipelines that returns normalized JSON for dozens of platforms, handles the anti-bot layer, and bills pay-on-success — you only pay when it actually returns your data. The same approach powers our anti-bot adoption study and the Google-vs-Bing-vs-Brave SERP study.
Read the web that blocks bots, as clean JSON
Crawlora turns Product Hunt, search engines, marketplaces, and social platforms into normalized JSON — handling proxies, rendering, and anti-bot — and bills pay-on-success. 2,000 free credits a month, no card.
Frequently asked questions
When did Product Hunt launch?
Product Hunt launched on November 6, 2013, built by Ryan Hoover as a Thanksgiving-break email list. Its leaderboard is therefore about a 12.5-year record (2014–2026) — and the entire AI wave is compressed into the back third of it.
How have Product Hunt trends changed over time?
The board moved through clear eras: generic mobile apps (2013–14), startup and maker resource directories (2015–16), a design-tools and productivity golden age (2017–19), a remote-work and COVID pivot (2020), a no-code peak (2021), the AI spark and breakout (2022–23), and an AI-agent monoculture (2024–26).
When did AI take over Product Hunt?
2023 is the inflection year — the first year 'Artificial Intelligence' is the single most common topic in the annual Top 10. AI products in the Top 10 grew 0 → 3 → 5 → 7 → 9 across 2021 to 2025, sparked by ChatGPT's debut in late 2022.
Do Product Hunt trends change month to month?
At the product level, yes — in 2026 the monthly Top 8 turns over almost completely each month, with essentially no repeat products. But the theme is stable: nearly every entry is an AI agent. What rotates month to month is the sub-flavor — coding, then design, then meetings, then commerce, then fundraising.
What is trending on Product Hunt in 2026?
AI agents. Almost every top launch in 2026 is an autonomous agent that does a job for you rather than a tool you operate, so the dominant theme barely moves — but the sub-flavor rotates roughly monthly: agent runtimes and code review, then AI design and agent marketplaces, then meeting-notes and evals, then agent-commerce and voice agents, then fundraising and GTM agents, plus a rising MCP-native tool-layer cohort that turns any site or app into an agent tool.
How was this Product Hunt trend analysis done?
We read Product Hunt's annual and monthly leaderboards directly through a structured web-data API and cross-checked platform facts against public sources. The annual board is a curated top-~20, scores are cumulative upvotes, and 2026 is year-to-date, so the figures are directional rather than exact.
Do Product Hunt winners survive?
Not reliably. Re-checking 28 era-defining Product Hunt winners in June 2026, about 7 in 10 still run the product that won, but roughly 3 in 10 are dead or have pivoted away — and a #1 badge was no guarantee (2013's Sunrise and 2024's #1 Wordware are both effectively gone). The deaths cluster in the 2020–22 cohort (~50%), the pandemic and social-novelty bets; trend-independent utilities and infrastructure survive best.