Công Vũ Đăng

@matarloum91 Fair point. Ember does not define relationship quality or score closeness. You choose who matters, what you want to remember, and when a reminder makes sense. The app's job is simply to help you hold onto the small context and show up intentionally; the relationship itself remains yours.

@mrqwenty Thank you, Matteo, and glad you grabbed the code. For bigger families, Ember supports groups so the people and context do not have to live in one flat list, and Keeper can work with group context too. The real test is whether it surfaces the right detail without adding noise. If you try it with a larger family, I would genuinely love to hear where it feels crowded or unclear.

@pzx553 Great question. Synced data lives in Supabase in Singapore, encrypted at rest and isolated per user with Row Level Security. When Keeper classifies a note or answers a question, only the context needed for that task is sent to an AI provider. In Settings > Keeper AI > Memory, users can view, edit, or delete every learned fact and insight separately; original notes can also be edited or deleted independently. That control is essential to the privacy promise.

@picknmixsweets Thanks for highlighting the core four, Olivia. If you had to choose one to test first, would it be the one-sentence capture, the reminders, privacy controls, or contact import? That answer would help us understand what lands most clearly.

@ray729coding Thank you, Ray. Privacy is the hard part, and the few sentences are not magic. Synced data is stored with Supabase in Singapore, encrypted at rest and isolated per user with Row Level Security. For an AI request, Keeper sends only the context needed to answer it. Ember cannot create closeness for you; it simply helps you remember the detail and the moment, then you decide to reach out.

@tooboo Thank you, Serj. We are deliberately keeping the core loop narrow: capture something quickly, surface it at the right moment, and make it easy to reach out. Scaling for us means making that loop deeper and more reliable before adding a long feature list. What capability would you expect next?

@mehulytwithai Thank you, Mehul! That is exactly the kind of person we built Ember for: someone who cares, but does not want small details to slip away. If you try it, I would love to hear which part helps most: quick notes, Keeper recall, or reminders.

Hey StartupBase! I built Ember because I kept forgetting the small things people told me - a friend's surgery date, what my nephew is into now, who just changed jobs. Contact apps store phone numbers; nothing helped me remember the human stuff.

With Ember you just type one sentence and the AI (we call it Keeper) files it under the right person. Ask later "what should I ask Mai about?" and it answers from everything you saved. Gentle reminders nudge you before birthdays or when you have been out of touch too long.

It is free to start on iOS and Android, and StartupBase folks get 1 month of Pro free with code SBASE-EMBER. Would love your feedback - especially on what would make you trust an app with this kind of personal data.

Geo link targeting plus UTM builder in the same tool is a smart combo - running region-specific store links (App Store vs Play Store by country) usually means chaining a shortener with a separate redirect service, so having it in one place with analytics attached saves real setup time. One question: on the privacy-first tracking, does that mean cookieless attribution? Curious how accurate the geo and device breakdowns stay without cookies.

The natural-language trigger examples are what stand out here - "update Salesforce based on recent customer emails" is exactly the kind of glue work that eats an hour a day and never gets automated, because setting up a traditional workflow builder takes longer than just doing the task. How do you handle mistakes though? If the AI mis-parses an email and writes wrong data into the CRM, is there a review step or an undo?

Consolidating feedback, support, and roadmap into one tool is a real pain point - most teams juggle a changelog tool, a support inbox, and a spreadsheet for feature requests, and things fall through the cracks between them. Curious how the roadmapping side handles duplicate requests: does it auto-merge similar feedback or is that manual? That is usually where these tools get messy at scale.