I'm Timon, the solo builder behind Loop. I wanted the agents I use in one desktop workspace, with their conversations, sessions and project knowledge close together.

The Brain holds project files, notes and saved decisions the agents can retrieve. Each agent keeps its own conversation. Freeforms lets me arrange chats and widgets around the task.

The waitlist is now open: https://loopagents.app/. Sign up to hear when early access opens.

The next demo I want to record would show one agent changing a project decision, then a second agent retrieving it and continuing without a pasted recap. Which task would you use to test that handoff?

I'm Henrique, one of the makers of Skyportal.

Instead of digging through dashboards, you ask it things like "Why did GPU utilization suddenly drop?" or "Which deployment caused this regression?" It looks through your deployments, Kubernetes events, GPU metrics and logs for the answer, then proposes a fix.

By default, each command that would change a server waits for your approval.

The client is MIT licensed. Install it with pip install skyportalai (Python 3.11+) and sign in with a Skyportal account.

I work on Asterion by KraftAI. We re testing a proposed $10 one-time report for solo founders and small agencies considering B2B software or browser tools.

We researched our own proposal and published this worked sample:

https://gist.github.com/hritikch...

Hey Product Hunt

Our website resonics.ai is live now!!

We re building a conversation memory layer for meetings and calls - something that helps individuals and teams remember what was decided, who owns what, what needs follow-up, and where the original context came from.

A lot of tools can record or summarise conversations, but the real pain usually comes later: finding the right decision, remembering the next step, or knowing why something was agreed.

What if a customer wants to try the product before signing up? Usually, they are forced to watch a static demo video.
The workforce management platform - HoursOps.com has built an interactive UX demo to make customers play with the screen before they proceed to signup.

Instead of asking customer to watch an outdated recording or create an account just to see the interface, we built an in-browser interactive demo.
Prospective users can test-drive the actual feature without a signup and interact with the features
Explore granular roles and workspace permissions
Create projects and manage task assignments
Approve timesheets and test the Work-to-Billing (Invoice) flow
Simulate isolated vendor sub-accounts
Try AI Assistant workflows
No sign-up. No credit card. Fictional data stays in the browser.
Test the full interface before deciding if it fits your team.
Try it now - https://hourops.com/demo

We build PikFind: guests take one selfie and get only the event photos they're in no app, no account.

Last week we ran our biggest event so far, a global pharma symposium (the client asked to stay anonymous): 21,688 photos, 439 attendees searched, 96% of selfie searches found the person, and 26,373 photos were downloaded.

Three things we learned:

1. International attendees are the real audience. 98% of visits used the English page a shared-folder link is useless to someone who has already flown home.

2. Real cameras produce weird files. A hidden null byte in the EXIF copyright field and 44-megapixel originals both broke our pipeline mid-event. We fixed them within hours, and now uploads retry and re-send on their own.

3. Organizers ask about privacy as much as speed. Search selfies are deleted right after processing, and attendees' removal requests (15 photos) were handled inside the event.

Hey everyone,

Yesterday we launched ActionMind on Product Hunt, and we're genuinely grateful to everyone who stopped by, upvoted, asked questions and shared honest feedback.

After nearly 20 years working in hospitality and service operations, I noticed something interesting:

Most businesses don t have a standards problem.

Rename update 4 November 2026: Open Typora is now Open Markdown Lane. It is the same independent open-source project, with version 0.23.0 and renamed desktop downloads.
I'm the developer. Open Markdown Lane is an MIT-licensed desktop Markdown editor that works with ordinary .md files in your folders: live editing, a source view, editable tables, split panes, and HTML/PDF export. The desktop interface now supports English and Simplified Chinese, with builds for macOS 14+ on Apple Silicon, Windows 10/11 x64, and Linux x64.
The optional AI workflow starts with a selection and your own improvement request. Generate a proposal, review each change, then apply it as one undoable edit. You can also explicitly choose up to four documents, generate an edit plan, and review the changes file by file.
AI uses your own model service and key. Selected text, instructions, and reference excerpts you explicitly include are sent to that service; costs, privacy, and offline availability depend on your configuration. Numbers, links, names, and Markdown syntax still need checking.
Source and downloads: https://github.com/AniChikage/op...
Basic editing demo: https://open-markdown-lane.comic...
The browser demo keeps edits in the current tab's memory; refreshing clears them, so download a copy first. Desktop folders, recovery, export, plugins, and AI are not part of the browser demo.
For technical notes and documentation, when do small selections give an AI editor too little context? Concrete examples would help me improve the workflow.
Disclosure: the project and this post use AI assistance. Open Markdown Lane, formerly Open Typora, is independent and unofficial, unaffiliated with Typora. It is still in early development and does not promise complete feature or behavior parity.