Connect with the Product Hunt community

A recent audit of 212 brands found that the average AI search share of voice dropped from 23.4% during discovery to 15.7% when buyers asked pricing-related questions.

The same audit found that more than half of the brands had zero visibility during pricing searches in their categories.

Six months ago, we were desperate for cash flow to keep the lights on. We decided to run a Lifetime Deal (LTD) campaign. We thought it was a win-win: we get a quick cash injection, and we get hundreds of users to give us feedback.

It worked... at first. We made $5,000 in a single week. I felt like a genius.

But the reality check hit us hard over the next 90 days:

• Our support tickets exploded by 400%. LTD buyers were incredibly demanding, asking for custom features and 1-on-1 onboarding calls for a product they paid $49 for once.

• Our server costs tripled. Many of these users were uploading massive files and using our API heavily. We were now paying AWS $300/month to host users who would never give us another dime in recurring revenue.

• Our real MRR stalled. We got so bogged down supporting the LTD crowd that we completely stopped our outbound marketing and sales efforts.

We hit #1 on Product Hunt on Launch Day!

Ended at #2 Product of the Day!

Then the Product of the Weekend!

Then #5 Launch of the Week

Hey everyone, Ola here

Specsight hit #1 Product of the Day on Saturday and I m still a bit in shock. Thank you for every upvote, comment and review, it means a lot as a solo founder!

When an AI application misses something, the tempting fix is to retrieve more. More chunks, more documents, maybe the entire attachment. Congratulations, you ve turned your context window into an expensive filing cabinet.

A Graphwise experiment illustrates the alternative. Across 11 questions over a dataset covering 6,000+ companies, graph-grounded retrieval used roughly 75,000 context tokens versus over four million for context stuffing. About a 54 reduction. Their write-up

That s a focused experiment, not a universal benchmark or a claim of 54 lower total costs. Building and maintaining the graph costs something too.

But the architectural lesson is useful. Relationships resolved during ingestion don t need to be rediscovered from raw text on every request. Retrieval can follow explicit connections and return targeted evidence rather than asking the model to reconstruct the connections itself.

Today, I came across a LinkedIn post that reported rough data on how many people were laid off across the biggest tech companies due to AI:

- canva: 1,271
- microsoft: 5,300
- meta: 10,400
- uber: 3,300
- atlassian: 1,380
- ebay: 1,100
- netflix: 1000
- hubspot: 600
- squarespace: 300
- dropbox: 200
total humans affected: 24,851

I understand the need for the founders to make the processes as effective (AKA layoffs), but it somehow reduces the economic power and ability to buy products (in the long term)

Two days out. The Google Cloud Run Hackathon is this Wednesday, October 14, and submissions close at midnight PT tomorrow.

The mechanics, one more time:
- Deploy and run your product on Google Cloud Run that's what makes you eligible
- Schedule your launch for Wednesday, October 14
- Get it in before midnight PT tomorrow

Every eligible launch picks up $50 in Google Cloud credits. Winners take $500 in credits, a spotlight on Google Cloud's own channels, and a 1:1 with the Cloud Run team.

People still ask what to use instead of Stay, and many old threads about it are out of date: Stay is free now. I wrote down where it stands today, the fixes that need no app, and what Display Maid, Moom, Rectangle Pro, Putback, Workflo and Snapset do according to their own sites. I make Many Displays, so the page also says where my app may not suit you.
https://manydisplays.com/stay-al...
If you have moved from one of these tools to another, I would like to hear about your setup and what made you change.

Hey everyone!
Thank you for all the support on Chopstick, it's been amazing to see all the love for it! We finished the day on #7!
I wanted to share the updates that have gone out since the release:

• Search has improved a lot. With one shortcut you can convert units, do math, view the weather, kill a port or do anything else a widget provides.

• Convert, compress or pass files to your agent by just dragging them onto the notch.

• Multiple new built in widgets including Clipboard History, Caffeinate and Activity Monitor.

• You can view built in and community widgets on https://chopstick.sh/widgets.

• Support for GitHub Copilot.

For a quick ask: I would love to see how you set up and use Chopstick. If you do a 20-minute call with me, I'll walk you through setup and listen to your feedback. In return I'll hand you a free license! If you already bought one, I'll refund you.

Join the discord here and send a quick message to #feedback-calls! I'll keep it to the first 10 people.

we've been testing websites for agent readiness while building Dedica, and one thing we're noticing is how many sites still struggle with basic agent interactions.

curious if anyone here is actually tracking AI agent traffic yet?

Lancetta 0.68.0 is out

Every Claude Code and Codex session, in one place: what waits on you, what you finished, and what the agents still keep on disk

You call a company, explain your issue to the bot, get transferred, and explain it again. Then you message them on WhatsApp the next day and start from zero.

The fix isn't a smarter model. It's shared memory: past interactions, decisions and customer context stored in one place, available to every agent and every human with permission to see it. When a customer says "can we pick up where we left off?", the next conversation should already know where that was.

This turned out to be one of the hardest parts of our platform to get right, especially with permissions.

Have you seen a company do this well? Which product made you feel remembered?
Support our launch - https://www.producthunt.com/prod...

What do you think about support - or have you even thought about it? =D

Do you think every product must have some kind of support?

Should it be a bot?
A human?
Email?
An AI agent?
Or all together?

In our case, with a complex workspace, I think working support is definitely a must.

This is the question we spend the most time on with every new customer at UnleashX.

One sales team wanted their AI employee to offer discounts on its own to close deals faster. Sounded fine until we asked, up to how much? Nobody had an answer. Their human reps had never had a written rule, they just knew from experience.

We ended up setting a limit: up to 40% on its own, anything above that goes to a manager. Deals closed faster, and managers only saw the ones that actually needed them.

Turns out giving AI autonomy forces you to write down rules your team has been following without ever saying them out loud.

Hi everyone, Dmytro here, CTO and co-founder of Easy-Peasy.AI.

AI Employees goes live on Product Hunt tomorrow, Thursday Nov 5, at 12:01 AM Pacific. You hire AI teammates by role (19 ready-made roles, or describe your own) and brief them in a group chat. The lead splits the work, teammates reply in a thread, and you get one write-up. Each employee has a memory, its own computer and routines that run on a schedule or when an email arrives. Emails, calendar changes and app actions wait for your OK.

I've been spending more time exploring what other indie developers and founders are building, and honestly, some of the most interesting products I've found aren't the ones getting thousands of upvotes.

I'll start!

We're building Dots, a macOS app that turns your Mac's notch into a customizable workspace. Instead of just music controls and notifications, you can bring in your own apps, websites, shortcuts, and even developer tools.

Netfox 0.53.0 is out

Settings now moves the way System Settings does

Next in our short tutorials, one task each.

This one: a chat on your own site. Pick the name and colour, choose who answers (an AI bot that hands over to your team, or just your team), and paste one snippet. Chats show up in the same inbox as calls, texts and email, with the page the visitor is on.

Video: https://youtu.be/XI9AV-yhn34

What should we cover next?

PasteKit currently handles 58+ formats and 48 conversion pairs, all running locally in your browser. Nothing is uploaded, and it works offline after your first visit. I want the next addition to come from you. What format or conversion do you reach for that's missing, or that you still paste into a tool you don't fully trust? Config formats, query languages, data formats, anything goes. Drop it below and upvote the ones you'd use too.

Hey everyone! We recently announced Asymptote Labs' $5.5M seed round and launched Beacon, our open-source, self-improving memory for AI coding agents.

Yesterday, Beacon also hit #1 Product of the Day on Product Hunt!

i've been talking to a lot of researchers lately and the same thing keeps

coming up. nobody is slow because the ideas are hard. they're slow

because the stack around the ideas is held together with tape.

Single agents are manageable. The moment agents start handing work to each other, failures stop being traceable. One bad output becomes the next agent s trusted input, and the break shows up far from where it started.
If you ve got a multi-agent system running reliably in production, what made it work? Narrower tool scopes, strict handoff rules, better tracing, fewer agents?

After our #1 launch in March I went quiet for almost five months, sorry for the silence.
When I came back, my old way of building with AI stopped working.

My Claude and Codex limits were gone in 2-3 days. Codex kept forgetting what the other sessions had done. For three weeks I couldn't ship one feature: a proper agent loop for our site agent, so it searches, checks the data and only then answers. I kept forwarding messages between the two AIs by hand, like a courier.

In the end I set up a harness: one shared space where every AI session, in any tool and on any computer, starts by knowing what we decided, what we tried and what's next. Within two days the stuck feature finally worked. Now my limits last the whole week.

I wanted to share this experience and discuss it here. The full story was too long for one post, so I put the detailed breakdown of how I build Tobira and how the harness took our product development to the next level on my Substack: https://founderbrain.substack.co...

Archived repositories stop counting as work in octoscope 0.40.0

I believe that visualizing our dreams and goals can make them come true.

We set a point where we want to be, and naturally start drawing lines toward it. We create sub-goals, sub-dreams =D and step by step, finally get where we wanted to be.