Two days ago I launched CalNote AI, an iPhone app that starts recording when a meeting on your calendar starts and saves a clean note to Files, Obsidian or Notion. I'm a product owner, not a developer, and this was my first launch anywhere.

The numbers, since I couldn't find many posts that share small ones:

• Final position: #500 for the day, 1 upvote, no outside comments

• New devices on launch day: 4 (a normal day is about 3)

• Total devices since Oct 20: 30 before the launch, 37 now

What I think went wrong:

I built Oreja after getting tired of reaching for Apple Music every time a ChatGPT voice reply started. It watches the selected app's output level locally, pauses Music while the reply is active, then brings it back after the quiet interval. There is also a lower-volume mode.

The whole thing is intentionally small. No OpenAI API, account, recording, transcript, analytics, or network service. The repository is open source, and the audio path feeds transient levels into a deterministic coordinator.

A delivery estimate needs room for briefing, client questions, feedback rounds and handoff. Here is a capacity check that a solo service provider can use before agreeing to another request.
Start with the hours available this week. Subtract business admin, an interruption allowance and work already promised to clients. Then compare a working estimate for the new request with a slower estimate.
Illustrative example. Start with 32 working hours. Subtract 7 for admin, 5 for interruptions and 8 for booked client work. That leaves 12 hours. A request estimated at 5 hours leaves 7. If it takes 7.5 hours, it leaves 4.5.
The useful decision is what that smaller remainder has to cover. List the other deadlines due that week, including review and handoff. If 4.5 hours leaves too little room, reduce the first delivery or agree a later start. Keep the interruption allowance in both estimates so that a slower project does not silently consume it.
I m a maker of SoloHatch. This is the method behind our free browser calculator, which lets you compare the two estimates without signing up.
https://solohatch.com/client-cap...
The companion guide includes a weekly worksheet and the worked example.
https://solohatch.com/solo-workf...
For feedback rounds, do you budget a fixed number of hours, an allowance per round, or a range? What helps you keep that estimate close to the actual work?

It's time companies moved from a passive marketing tech stack to a proactive one, so we're identifying the newest, most innovative tools out there.
What we're building: @Tulu , an autonomous operations & analytics platform that ties every marketing activity directly to closed-won revenue in HubSpot.

• Who it's for: LeadGen SMBs who are done guessing which campaigns drove revenue, and can't afford to pay someone to be a full-time data janitor.

• The hardest part: The "shit in, faster shit out" paradox. AI agents are great, but they fail if the underlying CRM data is broken.
  Now you! Since we're fixing the data foundation, I m curious:
  What are the hottest, most innovative AI marketing tools you're seeing right now?

A few years ago I started managing my career like a project. I scanned job ads for the role I wanted, listed the skills employers kept asking for that I didn t have, made a plan to learn them, and built projects to show them on my CV. It helped me land roles at two Fortune 500 companies in the US and one in Europe.

I m now building Mettlespark to do that for anyone in tech. You pick a target role and add your CV, or type in your skills. It reads current job ads for that role, shows your gaps ranked by how often employers ask for them, builds a learning plan with materials and projects, then helps with a CV and interview prep for each job you apply to.

It s early and free while in early access: mettlespark.com

WMM: my AI agent does the marketing, I still sign off on what ships

I build apps and handle their marketing myself. WMM is a local marketing workspace my own AI agent operates: it researches, drafts and schedules the work inside permissions I set per operation. The actions I choose to gate wait for my explicit approval of the exact proposal before anything runs, and when a reviewed action executes it leaves a saved receipt of what actually ran.

Honest state: public repo, one documented trial on my own app's Facebook Page - I approved the content, a scheduled run published it, and the readback from Meta matched what I approved. One post. Nobody is retiring off this yet.

What I want from this crowd: if your agent queued a week of marketing work for your review, what do you need to see per item before you'd approve - content, reason, destination, cost? And what would you never take out of the queue?

Hi Product Hunt!

As a Japanese language learner, I always felt that memorizing vocabulary and Kanji through traditional flashcards could get pretty tedious and boring.

That s why I built Nihongo Drop an interactive, browser-based typing game designed to make learning Japanese vocabulary fun, engaging, and effective.

Key Features:

The idea is fairly simple: the system is built around the dynamics of a non-equilibrium medium. There are no hard thresholds or conventional if-else rules, no forecasting algorithms and no AI. It responds to market shocks through mathematical conservation laws, with its internal state adapting to changing volatility. It's still an experimental system, not yet a prototype, but the simulation of physical states is already showing some interesting results. I'm approaching the problem through concepts borrowed from quantum mechanics, thermodynamics and statistical physics. These fields deal with complex systems where tracking every individual element and predicting its exact future state isn't always practical. Rather than trying to predict market direction, the engine continuously processes incoming exchange data through an evolving physical model. Each component affects the others, and the overall behavior emerges from their interaction. There's no learning from historical trades, searching for profitable patterns or memorizing winning combinations. What interests me most at this stage isn't individual trades, but how the system behaves as a whole during a session. The main thing I'm watching is how it handles energy and accumulated pressure. One encouraging change is that drawdowns have become considerably smaller than in earlier experiments, even under highly unstable market conditions. That's an important milestone. It suggests that the underlying mechanics may actually be working. The goal isn't to guess which way the market moves. It's to build a system that can absorb external pressure, redistribute energy, neutralize negative charges and eventually bring session equity above its previous all-time high. Video showing the draft in action (transactions and data can be found via the link in the video description): https://youtu.be/9FfTlnbw_xE

P.S. And yes, I require funding and a bit of assistance (data collection).

Hello everyone!

We're excited to introduce Nudge to the Product Hunt community!

I'm a computer science student exploring an idea for a personal finance app, and I want to validate it before investing time in building it.

One problem with expense trackers is that logging every transaction can feel tedious. Filling out forms, selecting categories, and entering details can make people stop tracking their expenses altogether.