Shopify
4/5Rating only
@aguna-health
Member since
Loading reviews...
Rating only
We set up three product launches on Product Hunt this week, one per product on different days. Scheduling a launch for a set day, adding makers, a gallery and a first comment is straightforward, and the pre-launch page gives you a badge for your own site straight away. One thing to plan for: the listing name has to be just the product name, not a tagline, so keep the description for the tagline field.
We check Perplexity every day to see which apps and tools it recommends in our categories, because more people now ask an AI before they search. It always shows the pages it pulled from, which is the useful part: you can see exactly which articles and lists shape the answer, and work on being in those. Answers can change from one day to the next, so we read them on a clean session and log them rather than trusting one result.
aWe are a very small company and Claude is our operations team. Projects keep all our docs in one place so every chat starts with the full context, and it drafts support replies, outreach and website copy, reads our inboxes and checks listings through the browser. It is strongest when you give it written rules and source pages to check facts against. It still needs those rules repeated in a project file, or it can answer from memory instead of checking first.
Our founder is not a developer, and Claude Code does most of the engineering on our iPhone and Android app: fixes, automated checks for every bug that gets reported, and builds. It reads the codebase before changing it and writes a test that fails on the old code, which is what makes it trustworthy. The weak spot is memory across long sessions: rules have to be written into a project file it reads every time, or it repeats old mistakes. Once that file exists it is a huge time saver.
We run three websites and use Search Console on all of them every week. URL Inspection tells you straight away whether a page is on Google, and Request indexing gets a new or rewritten page into the crawl queue fast. The sitemap report and the structured data warnings caught problems we would have missed. It is free, which matters for a small company. The one gap: it only covers Google, so we pair it with IndexNow for Bing.
Our whole app lives on GitHub, and GitHub Actions builds it: the Android release bundle and the iPhone build to TestFlight, plus a few hundred automated checks before anything ships. The workflow files are simple once you have one working. What we learned the hard way: the free minutes on a private repository run out mid-month, and then jobs are refused within seconds. A small public builder repository that checks out the private code with a token gets free runners and kept our releases going.
Higgsfield is where we make our app's mascot, character art and short video takes. Having many image and video models in one place, with a credit balance, saves jumping between sites. Lessons from real use: the video models animate from the very first frame, so every prop has to already be in the start image, because a prompt will not add or remove one. Plan credits per batch, since long video takes use them up fast. The output is strong enough for social ads.
We use ElevenLabs for every voice in our app's promo videos and for the coach voice. The newest voice model sounds like a real person, and asking for an excited delivery actually changes the read. Speech to text gives word-level timings, which we use to make the screen change on the exact word. Music generation gave us tracks made for each video. Watch which workspace you are in: credits are per workspace, and our API connection pointed at a different one from the web app.
We schedule our app's short videos to TikTok, YouTube and Instagram from one Buffer queue, and it saves us hours a week. Uploading a video once and posting it to every channel works well. Things to know: YouTube caps posts through Buffer at 10 a day, TikTok can throttle you for a day if you push too many posts through the API, and Instagram trial reels fail at publish time if the account is too small, even though Buffer shows them as scheduled. Check the error list, not just the calendar.
