Skip to main content

​
Your Weekly Launch Radar

The best products ship every day on Product Hunt — but nobody has time to check every day, read every comment thread, research every founder, and figure out which launches actually matter. You open the site on Friday and the week’s winners are already buried under today’s. That’s this use case. You give Wolffish one prompt and it does the full job: browses Product Hunt through your logged-in browser extension, collects every top launch from the past week, then goes deep on each one — the company behind it, the founders, any funding history, the landing page, a YouTube demo if one exists, the community’s verdict from the comment section, and the raw stats. The result is one clean, ranked, categorized PDF with an executive summary up top, every product numbered by rank with full stats, and just enough commentary to know what’s worth your attention. The point is that you shouldn’t have to scroll Product Hunt at all. Open the PDF, read the executive summary, skim the categories, and you know exactly what launched, who built it, whether it’s funded, and what the community thinks — all in one place.
This is a read-and-browse workflow — Product Hunt browsing via the extension, web search for company research, and file generation. It doesn’t post, vote, comment, or modify anything on your behalf. That also makes it a solid heartbeat candidate: a fresh launch briefing waiting for you every Monday morning.

​
What Makes This a Hunter, Not a Scroll

Three things separate this from skimming the Product Hunt homepage:

Full-depth research on every launch

It doesn’t just list the product — it finds the company behind it, the founders and their backgrounds, any funding rounds (past or new), the landing page, and a YouTube demo link if one exists. You get the full picture, not just a tagline.

Community consensus, not just upvotes

Upvotes tell you popularity. This tells you why — it reads the actual comment threads, extracts the general sentiment, and pulls the top 5 most valuable comments for each product so you hear the real signal.

Ranked, categorized, and numbered

Every product is numbered by rank, slotted into its category (AI, Developer Tools, Design, Productivity, etc.), and laid out with full launch stats — upvotes, comments, makers. No guesswork about what won the week.
It all lands in one PDF, designed to be read in ten minutes, not a wall of product cards you’ll never finish.

​
Capabilities Required

  • computer-use — the primary data source. Wolffish uses the browser extension to browse Product Hunt through your already-logged-in session — navigating the leaderboard, opening each product page, reading descriptions, stats, and comment threads. Product Hunt is already linked and accessible via the extension.
  • web-search — the research engine. For each product, web_search and web_fetch find the company website, founder bios, Crunchbase/LinkedIn profiles, funding history, and YouTube demo videos. This is where the depth comes from.
  • shell — renders the finished briefing. The agent writes styled HTML and prints it to PDF through a headless browser (Playwright/Puppeteer via npx) — the same pipeline behind PDFs Everywhere and CV Tailor.

​
Setup

Read the general Setting Up for Success guide first — what’s below is specific to this workflow. What matters here is agentic reliability and research depth. The hard part is navigating Product Hunt cleanly, then running a thorough research sweep on each product — finding the real company, the actual founders, separating genuine funding announcements from press speculation, and distilling dozens of comments into a useful consensus. You want a model that’s methodical.
  • DeepSeek V4 Pro on Max reasoning mode — the recommendation. Frontier-class agentic tool use and reasoning at a fraction of the cost. Max mode gives it the full reasoning budget for the research calls — finding the real Crunchbase page, not a similarly-named company, and distinguishing a seed round from a press rumor. Set the reasoning effort to Max in Settings > Models. See DeepSeek configuration.
  • Claude Opus / Sonnet 4.x — strong alternatives if you’re on Anthropic; Opus is the most thorough at the founder/funding research stage, at 10–20× the price.
  • Avoid low-effort configs — DeepSeek Flash, any model on None reasoning mode, or Haiku. They can browse Product Hunt fine, but the company research and comment synthesis is where a model that doesn’t deliberate will hallucinate funding rounds or conflate similarly-named companies.

​
Required

  • Wolffish installed and running — the desktop app with a configured brain workspace.
  • A capable cloud API key — DeepSeek (V4 Pro) recommended, or Anthropic (Claude); configured in Settings > Models.
  • Product Hunt accessible via the browser extension — Wolffish’s browser extension must be installed and connected, with Product Hunt already accessible (logged in or publicly browsable). The extension is how Wolffish reads Product Hunt pages.
  • Brave Search API key — configured in Settings > Services > Brave Search. The company/founder/funding research fires many searches per product; the default search will rate-limit partway through a full week’s worth of top products. With a Brave key, the research runs clean. The free tier is plenty.

​
Optional

  • filesystem — for a cleaner audit trail when writing the HTML and saving the PDF (the shell pipeline works without it).
  • A headless browser — Wolffish installs Playwright/Puppeteer on first use if neither is present. Having npm/npx on your PATH makes that instant.

​
What the Hunter Tracks

For every top product from the past week, the hunter collects:
Data pointWhere it comes fromWhy it matters
Rank & upvotesProduct Hunt leaderboardWho won the week, numerically
Comments countProduct Hunt product pageEngagement signal beyond upvotes
CategoryProduct Hunt tags / descriptionOrganize the PDF by domain
Tagline & descriptionProduct Hunt product pageWhat it actually does
Website linkProduct Hunt product pageGo try it yourself
YouTube / demo videoProduct Hunt page + web searchSee it in action
Top 5 commentsProduct Hunt comment threadThe community’s real signal
General consensusSynthesized from commentsThe verdict in one line
Company & landing pageWeb searchWho’s behind it
Founders & biosWeb search (LinkedIn, Crunchbase, about pages)Track who’s building what
Funding historyWeb search (Crunchbase, TechCrunch, press)Bootstrapped vs. backed — and how much

​
The Prompt

Send this to Wolffish on-demand, or put it on your heartbeat. It’s long because “ranked, researched, and beautifully laid out” is a real contract — every block below earns its place. 
You are my Product Hunter. Your job: track every top
product launch on Product Hunt from the past week, research
the company and people behind each one, read the community
verdict, and hand me ONE clean, modern, ranked PDF I can
read in ten minutes. Miss nothing in the top launches;
cut nothing from the research.

FIRST, fix the window. Run `date +%F` to get today. The
window is the last 7 days through today.

=== STEP 1: BROWSE PRODUCT HUNT (use the browser) ===
Open producthunt.com in the browser via the extension.
You are already logged in — just navigate.

Go to the weekly leaderboard / top products. Collect the
top launches from the past 7 days. For EACH product on
the leaderboard, open its product page and grab:
- Product name and tagline
- Rank position (by upvotes for that day/week)
- Upvote count
- Comment count
- Maker(s) / hunter name(s)
- Category tags (AI, Developer Tools, Design, Productivity,
  SaaS, Fintech, Health, Education, etc.)
- Direct link to the Product Hunt page
- Website URL (from the product page)
- YouTube or demo video link if listed on the page
- The product description / pitch

Then open the COMMENTS section for each product and:
- Read through the comment thread
- Extract the TOP 5 most substantive comments (not "great
  product!" — real feedback, questions, or insights)
- Write a ONE-LINE general consensus: what does the
  community actually think? Positive, mixed, or negative,
  and WHY. Be honest — if the comments are thin or mostly
  from the makers, say so.

=== STEP 2: RESEARCH EACH PRODUCT (use web search) ===
For EACH product collected above, run web searches to find:

COMPANY:
- Official company name (may differ from product name)
- Company website / landing page (visit it with web_fetch)
- What the company does beyond this product (if anything)
- Headquarters / location
- Year founded
- Company size (approximate employees)

FOUNDERS:
- Founder name(s) and role(s) (CEO, CTO, etc.)
- Brief background — previous companies, notable roles,
  education if readily available
- LinkedIn profile link(s) if findable
- Twitter/X handle(s) if findable

FUNDING:
- Search for ANY funding — past rounds, new announcements,
  or nothing at all
- If funded: round type (pre-seed, seed, Series A, etc.),
  amount, date, lead investors
- If bootstrapped or no funding found, say so clearly
- Source the funding info (Crunchbase, TechCrunch, press
  release, etc.)
- Check for RECENT funding news around the launch — a
  Product Hunt launch often coincides with an announcement

Use Crunchbase, LinkedIn, TechCrunch, company about pages,
press releases, AngelList / Wellfound, Twitter/X. Grab what
is readily available — do NOT rabbit-hole on any single
product. If something isn't findable, leave it blank and
move on.

=== STEP 3: BUILD THE PDF ===
Write it as styled HTML, then render to PDF with a headless
browser. Save to the workspace as
product-hunter-YYYY-MM-DD.pdf (today's date).

Design: clean, modern, minimal, COLOR-CODED by category,
skimmable in ten minutes. Let stats, badges, and layout
carry the load — not prose.

Structure, in this order:

1. MASTHEAD — "PRODUCT HUNTER" large, the date window
   ("Week of <start> – <today>"), total products tracked,
   and a one-line tagline.

2. EXECUTIVE SUMMARY — the 5 most important launches of
   the week, one tight line each with rank number, category
   badge, and upvote count. If I read only this, I know
   what won the week.

3. WEEK AT A GLANCE — a compact dashboard:
   - Total products tracked, total upvotes across all,
     total comments across all
   - Bar chart: products per category (inline SVG or CSS
     bars — NO external chart libraries or CDNs)
   - Top 5 by upvotes (horizontal bar chart)
   - Funding breakdown: how many funded vs bootstrapped vs
     unknown

4. PRODUCT CARDS — one per product, NUMBERED by rank
   (highest upvotes first). Each card contains:
   - Rank number (large), product name, tagline
   - Category badge (color-coded)
   - Stats row: upvotes | comments | maker(s)
   - One-paragraph description (from Product Hunt)
   - Links row: [Website] [Product Hunt] [YouTube/Demo]
     (if available) — as clickable URLs
   - COMMUNITY VERDICT: the one-line consensus + top 5
     comments (quoted, with commenter name)
   - COMPANY PROFILE: company name, website, HQ, founded,
     size, what they do
   - FOUNDER SPOTLIGHT: name(s), role(s), background,
     links
   - FUNDING STATUS: badge (Funded / Bootstrapped /
     Unknown) + details if funded (round, amount, date,
     investors). If recent funding news found, highlight it

   Use category colors to accent each card. Make cards
   visually distinct and scannable — someone should be able
   to flip through and get the gist of each product in
   15 seconds.

5. CATEGORY BREAKDOWN — one section per category that had
   products this week, listing which products fell under
   it with their rank numbers. Quick cross-reference.

6. METHODOLOGY — short: the exact window, how products
   were collected (Product Hunt leaderboard via browser
   extension), what was researched via web search, that
   funding info comes from public sources and may be
   incomplete, and that comment consensus is synthesized
   from the thread (not every comment).

Footer on every page: "Generated by Wolffish" left,
"Page X of Y" right.

=== WHEN DONE ===
Tell me three things: the #1 product of the week and why
it won, the most interesting company story (funding,
founder background, or pivot), and any product where the
community consensus was notably negative or skeptical —
so I know the full picture.
The prompt is designed to work as-is — the only thing you might change is the time window (the last 7 days). Shorten to last 24 hours for a daily pulse, or extend to last 30 days for a monthly deep-dive. Everything else — the research depth, the comment analysis, the PDF layout — stays exactly as-is.

​
Customize It

Change any of these to make the hunter yours:
WhatWhere in the promptIdeas
Time windowthe last 7 dayslast 24 hours for today’s launches only, last 30 days for a monthly round-up
Research depthThe company/founder/funding blocksStrip funding research for a faster run, or expand to include competitor analysis
Comment depthTOP 5 most substantive commentsIncrease to 10 for a fuller picture, or drop to 3 for speed
Category focusAdd a filter lineOnly track products tagged AI or Developer Tools to narrow the sweep
LanguageAdd a line at the endWrite the entire PDF in Arabic — the headless-browser pipeline renders RTL cleanly
Accent paletteThe color rulesMap categories to your brand colors

​
How It Works

The prompt drives a three-stage pipeline. Prefrontal loads the computer-use, web-search, and shell capabilities into context, the model fixes today’s date, and runs:
StageWhat happensOutput
1 — BrowseOpens Product Hunt via the browser extension, navigates the weekly leaderboard, opens each product page, reads descriptions, stats, and full comment threadsRaw product data: names, stats, comments, links
2 — ResearchFor each product, web_search and web_fetch find the company website, founders, LinkedIn/Twitter profiles, Crunchbase entries, funding history, and YouTube demosCompany profiles, founder bios, funding cards
3 — Design & renderStyled, category-color-coded HTML with executive summary, dashboard charts, ranked product cards, and category breakdown → printed to PDF via headless browserproduct-hunter-YYYY-MM-DD.pdf
Hippocampus logs the run as an episode and Basalganglia records the outcome, so each briefing is kept for reference — handy when you want to track a product you spotted three weeks ago or compare launch trends over time.

​
What’s in the PDF

The output is engineered to be read top-to-bottom in ten minutes, or skimmed by rank in two:
SectionWhat it gives youWhy it’s there
MastheadTitle + the exact week window + product countYou know precisely what period and how many products this covers
Executive SummaryThe week’s top 5 launches, one line each, rankedCaught up in 60 seconds if that’s all you read
Week at a GlanceProduct counts, category split, upvote leaders, funding breakdown + chartsThe shape of the week at a glance — no reading required
Product CardsEvery product: ranked, categorized, stats, links, comments, company, founders, fundingThe substance — everything you need to evaluate each launch
Category BreakdownProducts grouped by domain with rank cross-referencesFind what’s relevant to your field instantly
MethodologyWindow, collection method, research sources, caveatsKnow how the sausage was made — and its limits

​
Example Run

Here’s the shape of a finished briefing. This is an illustrative example of the layout and tone — your real run pulls live data from the actual week. Note how each product card gives you the full picture — stats, consensus, company, founders, and funding — in a scannable format.
PRODUCT HUNTER Week of 11–18 June 2026 — 24 top launches tracked
EXECUTIVE SUMMARY
  1. 🟦 AI — CodePilot 2.0 (2,847 ▲) — AI pair programmer that generates full test suites from natural language; community calls it “the first tool that actually saves time”
  2. 🟩 Developer Tools — StackForge (2,103 ▲) — One-click full-stack deployment platform; praised for zero-config approach
  3. 🟪 Design — PixelMind (1,892 ▲) — AI design system generator; mixed reception on output quality
  4. 🟧 Productivity — FlowState (1,654 ▲) — Focus timer with biometric feedback via Apple Watch; strong positive consensus
  5. 🟥 Fintech — SplitLedger (1,441 ▲) — Real-time expense splitting for teams; community flagged pricing concerns
WEEK AT A GLANCE
2418,4322,891
products trackedtotal upvotestotal comments
Products per category AI ████████ 8 Developer Tools ██████ 6 Design ████ 4 Productivity ███ 3 Fintech ██ 2 Health █ 1Funding status — 💰 Funded 9 · 🏗️ Bootstrapped 11 · ❓ Unknown 4
#1 — CodePilot 2.0 🟦 AI · 2,847 ▲ · 342 comments · Made by @sarah_chenAI pair programmer that generates complete test suites from natural language descriptions of expected behavior.🔗 Website · Product Hunt · YouTube DemoCommunity Verdict: Overwhelmingly positive — commenters highlight the test generation quality and time savings; the few concerns are about pricing for solo developers.Top Comments:
  1. “Finally a tool that actually understands what I want to test, not just autocomplete.” — @devmike
  2. “Used it on a real codebase for a week. Saved me ~3 hours per day on test writing.” — @janedoe
  3. “The natural language → test pipeline is genuinely better than Copilot for this specific use case.” — @testguru
  4. “Pricing seems steep for indie devs. Any plans for a free tier?” — @solofounder
  5. “Integration with CI/CD pipelines out of the box is a huge plus.” — @ops_lead
Company: CodePilot Inc. · San Francisco, CA · Founded 2024 · ~25 employees Founders: Sarah Chen (CEO) — ex-Google DeepMind, Stanford CS PhD · James Liu (CTO) — ex-Stripe, built their internal testing framework Funding: 💰 Series A — 18M(March2026),ledbySequoiaCapital.Previouslyraised18M (March 2026), led by Sequoia Capital. Previously raised 4M seed from Y Combinator (2024). (Source: Crunchbase, TechCrunch)
#3 — PixelMind 🟪 Design · 1,892 ▲ · 287 comments · Made by @alex_designAI-powered design system generator — describe your brand and get a complete component library.🔗 Website · Product HuntCommunity Verdict: Mixed — designers praise the concept but several note output quality doesn’t match hand-crafted systems; makers are actively responding to feedback.Top Comments:
  1. “Love the idea but the generated components feel generic. Needs more customization.” — @uxlead
  2. “Perfect for MVPs and hackathons. Not ready for production design systems.” — @designer_j
  3. “The color palette generation is actually excellent. Components need work.” — @brandsmith
  4. “Tried it on our brand. 60% usable out of the box, which is impressive for v1.” — @startup_cto
  5. “Founders are super responsive in comments — good sign for the product’s future.” — @phhunter
Company: PixelMind Labs · Berlin, Germany · Founded 2025 · ~8 employees Founders: Alex Müller (CEO) — ex-Figma, previously founded a design agency · Nina Petrova (CTO) — ex-Vercel Funding: 🏗️ Bootstrapped — no funding found. Company appears to be self-funded.
(…one card per tracked product, ranked by upvotes)
CATEGORY BREAKDOWN🟦 AI — #1 CodePilot 2.0, #6 SynthVoice, #8 DataWeave, #11 PromptKit… 🟩 Developer Tools — #2 StackForge, #7 GitRadar, #12 APIForge… 🟪 Design — #3 PixelMind, #9 FontFlow, #14 ColorSpace…
METHODOLOGYWindow: 11–18 June 2026 (last 7 days incl. today). 24 products collected from Product Hunt weekly leaderboard via the browser extension. Company, founder, and funding data researched via web search (Crunchbase, LinkedIn, TechCrunch, company websites, press releases). Funding information comes from public sources and may be incomplete or outdated. Comment consensus is synthesized from the full thread — not every comment is quoted.Generated by Wolffish — Page 1 of 8
The example above uses fictional products and companies — your real run pulls live data from the actual Product Hunt leaderboard. Notice the PixelMind entry: rather than spin a mixed reception as positive, the hunter reports the community’s honest verdict and notes the specific criticism. That honesty is the feature.

​
Limits

  • Product Hunt’s leaderboard is the source of truth for “top launches.” The hunter tracks what Product Hunt surfaces — if a product launched but didn’t make the leaderboard, it won’t appear. Very new launches (last hour) may not have settled into their final rank.
  • Company research depends on public availability. Some startups are stealth or pre-launch with minimal web presence — the hunter grabs what’s there and leaves blanks when information isn’t findable. Don’t assume blank = nonexistent.
  • Funding data is point-in-time. Crunchbase and press coverage may lag actual rounds. A “bootstrapped” label means no funding was found, not necessarily that none exists. Treat funding cards as directional.
  • Comment consensus is synthesized, not polled. The one-line verdict is the model’s reading of the thread — it captures the dominant sentiment but may miss nuance in very long or polarized discussions.
  • Browser extension access is required. Wolffish reads Product Hunt through the browser extension on your already-logged-in session. If the extension isn’t connected or Product Hunt isn’t accessible, the browse stage will fail.
  • PDF rendering needs a headless browser. Wolffish uses Playwright or Puppeteer; if neither is installed it will try to install one, which usually works but needs npm/npx on your PATH.

​
Cost & Model Guide

Heavier than a simple list — the per-product company research, founder lookups, and funding searches mean many searches and fetches. Still inexpensive on the recommended model.

​
Approximate Cost Per Run

ModelEst. Cost Per RunNotes
DeepSeek V4 Pro (Max mode)~$0.10–0.35Recommended. Strong agentic browsing + thorough research at a fraction of the price
DeepSeek V4 Flash~$0.04–0.12Cheaper and faster; shallower on founder/funding research — fine for a quick overview
Qwen 3.7 Max~$0.20–0.50Solid alternative
Claude Sonnet 4.x~$0.50–1.20Polished prose; thorough research
Claude Opus~$2.00–4.50Most thorough at the research stage; overkill for weekly use

​
Token Budget

~300,000–700,000 tokens per run. The Product Hunt browsing (opening each product page and reading comments) dominates; the per-product web research adds the rest. A configured Brave Search key keeps the research from stalling on rate limits. Roughly 40–80 LLM calls (browser navigation, page reads, searches, fetches, and the HTML render).

​
Automating with Heartbeat

This is a strong heartbeat candidate: it reads Product Hunt through the browser extension (no posting, no voting, no commenting) and writes a single file. The only consideration is that computer-use heartbeats require the browser extension to be connected when the heartbeat fires — make sure the extension stays linked. (For the general rule on what’s safe to automate, see What to Schedule.) Open Settings > Heartbeat, paste the block below, and a fresh launch briefing will be waiting in your workspace every Monday morning. 
## Product Hunter | Weekly (Monday 08:00)

You are my Product Hunter. Track every top product launch
on Product Hunt from the past 7 days, research each one,
and deliver ONE clean, ranked PDF I can read in ten
minutes.

Run `date +%F` for today; window is the last 7 days.

Browse producthunt.com via the browser extension (already
logged in). Collect the top launches from the weekly
leaderboard. For EACH product, open its page and grab:
name, tagline, rank, upvotes, comments, makers, category
tags, website URL, YouTube/demo link, description. Read
the comments — extract the top 5 substantive comments and
write a one-line consensus.

Then web search each product: company name + website +
HQ + founded + size; founders + roles + backgrounds +
LinkedIn/Twitter; funding (round, amount, date, investors)
or "bootstrapped" if none found. Source: Crunchbase,
LinkedIn, TechCrunch, company pages, press releases. Grab
what's there, don't rabbit-hole.

Build the PDF as styled HTML then render via headless
browser; save as product-hunter-YYYY-MM-DD.pdf. Sections:
masthead (title, window, product count); executive summary
(top 5, one line each, ranked); week-at-a-glance dashboard
with inline-SVG/CSS charts (products per category, top 5
by upvotes, funding breakdown — no external chart libs);
ranked product cards (rank number, name, tagline, category
badge, stats, links, community verdict + top 5 comments,
company profile, founder spotlight, funding status);
category breakdown; short methodology + caveats. Footer:
"Generated by Wolffish" / "Page X of Y".
Weekly (Monday 08:00) gives you a full week’s launches every Monday. For a daily pulse, use Daily (08:00) and change the window to “last 24 hours” — you’ll get yesterday’s top launches each morning. For a monthly deep-dive, use Monthly (1st 08:00) with a “last 30 days” window.

​
Make Your Own

The pattern generalizes to any “track what’s launching and who’s behind it” job:
  • Indie Hackers radar — track launches on Indie Hackers instead, focus on bootstrapped MRR stories and growth tactics.
  • App Store scout — top new apps on the App Store or Google Play, with the same company/founder/funding research depth.
  • Y Combinator batch tracker — each YC batch’s demo day launches, with funding and traction follow-ups.
  • Competitor launch watch — narrow the Product Hunt filter to your category and track only competitors and adjacent products.
The discipline stays the same every time: every top launch tracked, every company researched, every community verdict captured. Change the source; keep the hunter.