Connect with the Product Hunt community

Most teams scaling Voice AI think they can monitor quality with a simple LLM prompt. They are wrong.

An LLM can t hear a "crunchy" voice line, it can t accurately measure a 500ms "barge-in," and it struggles with the nuances of true conversational flow.

When we built Cekura Monitoring, we realized we had to go beyond the LLM. We combined Heuristic and Statistical models with our Metric Optimizer to solve the "Scaling Wall."

A few small yet critical updates as we are heads down behinds the scene preparing for something bigger!

Revamped Referrals: Share Recall, get up to 50% off.

We've completely revamped the Recall Referral Program. Here's how it works:

We spent the last 4 months tracking 473 pieces of AI-generated content across our own site and customer sites. 218 got cited by ChatGPT or Perplexity. 255 got ignored. 12 got flagged in reader feedback as "low quality" or "clearly AI."

We wanted to understand what separates the ones that work from the ones that don't. Here's what the data showed.

The content that got cited

Three things stood out.

A Complete Guide to AI dictation for Mac

Road to 1,000,000 #Votap users Day 59 | Current: 1288

I've been trying to make Termsy as useful as possible since the day we launched.

And Termsy is fundamentally and significantly better now. Here's why:

1. Higher Accuracy: Rebuilt from the ground up with a smarter algorithm. To be honest, looking back, I feel a little embarrassed by the old algorithm. It threw up way too many false positives and wasn't trained on enough real-world data. Over the last few weeks, I've been analyzing 200-300 ToS pages across categories to find patterns to improve the accuracy. I've been actively improving the algorithm, so it's only going to get better in the newer versions.

2. Introducing Proactive Nudges: To catch the fine print early. Two things happen now once you encounter a ToS page: 1) The icon in the extension bar lights up 2) A self-dismissing widget pops up that nudges you to investigate. I didn't want you to forget to check Termsy when you need it most. So now, it shows up to give you more clarity so that you don't agree to something blindly.

3. Introducing Right Click to Investigate: To check the fine print without opening the link. Brands have become smarter, instead of letting you read the agreement, they usually write "By signing up I've read the Privacy Policy and agree to the Terms and Conditions" and link their pages. An already long verbose ToS is now another click away, hidden away. With a simple right click, you can now scan the terms without navigating away from the page. This is an attempt to shift the balance of power slightly back toward the consumer ensuring critical information isn't buried behind pages and long walls of text.

4. Now needs more permissions: to power all the new features. To do all this, we need more permissions, primarly to show that self dismissing widget and to read the contents of the page you aren't on (for Right Click to Investigate). This means your extension may be turned off. Chrome does this after major updates. So just click the extension bar icon, go to "manage extensions" to re-enable Termsy and unlock the full power of the new version.

While I pride myself in Clarity being the first of it's kind, I'm also extremely excited that we are currently building the worlds first data set of verified caffeine products and spotlighting brands for actually knowing the caffeine content of their drinks.

There are some other 'caffeine trackers' in the wild, but they all pull their data from the same available data set which is unfortunately not accurate. Clarity will only feature brands and products that know their caffeine content, this allows you to make truly strategic decisions about your dosing and timing.

A lot of the conversation focuses on replacement, but we re also interested in where people think the market may be too pessimistic.
P.S. We are giving away free credits for early traders. Go try it now.

Radion has leveled up again and it s more powerful than ever
Version 5 introduces Nintendo game imports along with automatic progress syncing , making it easier than ever to keep your library up to date .
This release also includes many improvements and features shaped by your feedback
Check it out now: https://radion-app.com
What s new:
- Import your games from Nintendo Switch, Switch 2, Wii U, and 3DS
- Automatically sync your Nintendo game progress
- Edit or delete backlog games directly from the game details page
- View estimated playtime in the progress picker
- Access the comment badge on all backlog games
- Edit comments directly from the backlog game card
Improvements & fixes:
- Larger, easier-to-tap close button
- Fixed missing achievement translations
- Fixed expired session issues for Xbox, Steam, and PlayStation Radion will ask you to reconnect if needed
- Security enhancements
- Performance improvements
Thanks for using Radion!
Good Game

Hey Product Hunt! We launched Concipe earlier this month to help teams turn scattered feedback into buildable specs you can use with CLI Agents and AI Code editors such as Claude and Cursor.

Since then, one question kept coming up: How strong is the evidence behind an opportunity?

So we shipped a few updates to make that clearer.

Every opportunity now shows an Evidence Score from 0 to 100. You can see how many unique voices mentioned it, which sources it came from, sentiment breakdown, and which customer tier is asking.

We surveyed 3,000 B2B buyers who used AI assistants (ChatGPT, Perplexity, Gemini) in their last vendor selection process.

The goal was simple: find out exactly what they asked and what actually influenced their shortlist.

The results reveal a massive gap between how companies market themselves and how buyers actually discover them through AI.

Hey PH community,

Two days ago (March 18th), we launched Genie - Databox's AI analyst, and we finished #3 Product of the Day!

Most AI visibility tools show you a dashboard. They tell you where you're invisible. They show you competitors winning.

Then what?

You stare at the screen, knowing there's a gap, but no idea how to close it.

I ve been using a lot of AI-generated code lately, and while it definitely speeds things up, security feels like a weak spot.

I ve run into issues like missing auth, exposed endpoints, and weak configs stuff that AI doesn t really flag unless you explicitly ask.

Curious how others are handling this:

• Do you rely more on manual reviews or tools?

• Any workflows that consistently catch vulnerabilities?

• Have you faced any real incidents because of AI-generated code?

Meow world, welcome back to The Breakpoint, a weekly thread on all things dev tools on Product Hunt.

The latest

Recent dev-first products launched on the site

As builders, we love tracking daily metrics: MRR, GitHub commits, daily workouts, Inbox Zero. Standard habit trackers are incredibly optimized for this gamification.

But lately, I've realized my "mental RAM" gets completely eaten up by the irregular tasks. The stuff you only need to do every few weeks or months:

• Changing the AC filter

• Watering specific houseplants

• Following up with that one dormant enterprise lead

• Taking as-needed medication

• Taking a full day away from the screen

Hey Hunters!

When I launched BrainLoom (The Local-First Learning OS) two months ago, the goal was simple: Your second brain should live on your hard drive, not a server.

Hey makers! Big update for discli the Discord CLI for AI agents.

What's new in v0.7.0
We went from 35 commands to **63 commands** and from 31 serve actions to **54 actions**.

Here's the highlight reel:
Interactive Components: Buttons, select menus (string, user, role, channel pickers), and modal forms. All via the serve JSONL protocol. Your bot can now send a dropdown, collect a form response, or disable a button after it's clicked.
Rich Embeds: Full embed support with color, footer, image, thumbnail, author, and repeatable fields. Both CLI (`--embed-color ff0000 --embed-field "Key::Value::true"`) and serve mode (JSON).
AI Agent: A single Python script (`ai_serve_agent.py`) that wires Claude Agent SDK to discli serve. Claude has full Discord control it runs CLI commands for messages/channels/roles and outputs component blocks for buttons/selects/modals. ~550 lines for an agent that can do literally anything on Discord.
New command groups:
- `discli webhook list/create/delete`
- `discli event list/create/delete` (guild scheduled events)
- `discli member timeout` (modern Discord moderation)
- `discli channel edit/forum-post/set-permissions`
- `discli thread archive/rename/add-member/remove-member`
- `discli role edit`, `discli reaction users`, `discli poll results/end`
7 Claude Code Skills: Install with `npx skills add DevRohit06/discli@discord-bot`. Skills for bot scaffolding, AI agents, moderation, support bots, welcome flows, logging, and slash commands.
Install:
pip install discord-cli-agent
Try the AI agent:
pip install discord-cli-agent claude-agent-sdk
discli config set token YOUR_BOT_TOKEN
python examples/ai_serve_agent.py
Then just @mention your bot:

Hi everyone   

For a long time, the cybersecurity community has been stuck in a cycle that feels increasingly broken. We have incredible tools for Web App and API pentesting that can scan thousands of endpoints in minutes, yet security teams are more overwhelmed than ever.

The problem isn't a lack of data; it s a noise problem.
If you ve spent any time in AppSec, you know the drill: you run a scan, and you re handed a massive list of "potential" vulnerabilities. Most of these turn out to be theoretical risks or outright false positives.

So teams end up stuck between two imperfect options:
 Automated security tools that are fast but noisy, or manual pentesting that is deep but impossible to scale.

npm install -g intina47/jot

https://github.com/Intina47/jot

You probably don't need this.

You have Notion. You have Obsidian. You have approximately forty seven other apps that promised to fix your notes situation.

Shipping AI tools is getting weird.

We tried to release a VSCode extension.

It kept hitting 500 errors. No explanation.

Hi everyone!

A few days ago I spotted @rohanrecommends sharing PH s brand new Kitty Coin leaderboard. This is definitely one of the biggest changes on PH recently.

Now it s baked right into every profile homepage:

Hey folks, so we've had some time since launch and I'm genuinely curious how you're using ClickSay in your day-to-day work. Are you using it the way we imagined, or have you found totally different use cases?

I'm especially interested in the friction points - where does it feel clunky? What would make it actually save you time instead of just being another tool?

Velocity is breaking out of Figma! 'Vibe Check' benchmarks the live web, no plugins needed. Lovable, Make, Claude code all supported. A/B test a bunch of designs or competitors. Velocity has cultivated a great community of designers and PMs as a Figma plugin, but recently design has evolved outside of Figma alone. We are extending the options for improving your earliest prototypes by porting the same great time-to-value reporting to the web.

Calling all PMs and designers to discover the time-to-value on their most important journeys starting on Wed 25th March.

Hi everyone quick note from the ZeroThreat team  

As AI begins to enter security workflows, there s a growing conversation around AI-driven pentesting. 

But one thing becomes clear quickly: fully autonomous AI attacking applications can introduce governance, safety, and operational risks.