Talkmore

building @talkamoreai | prev eng @sendaifun

Talkamore is an AI journaling app that holds onto context across sessions, so the things you wrote last month still inform what it says this week. The product is aimed at people working through personal decisions late at night who do not have a neutral person to talk to, the kind of moment where a generic chatbot forgets you the next day. You write what is on your mind, ask what you keep getting stuck on, and the app surfaces the relevant prior thread (a name, a recurring pattern, something mentioned in passing) without making you re-explain your life. Conversations are rolled up into a daily journal entry and indexed, so the next time you write, a relevant past moment surfaces. Users can pick between four personas with distinct tones, including Maya as a warm observational default, Sage for slower reflective sessions, Theo for structural framing, and Luna for blunt feedback. Conversations are stored encrypted at rest using AES-256-GCM with per-user keys, and the company says they are not used as training data, shared, or sold. The launch matters now because long-context LLMs have made persistent personal memory technically workable, and Talkamore is one of the first consumer products built specifically around that single idea rather than treating memory as a settings toggle. It is built by Ishan Lakhwani, a San Francisco based engineer who previously worked at Sendai (sendaifun) on crypto mobile apps, and who has said publicly that he started the project after dumping four years of his own journals into ChatGPT during a breakup and realizing the assistant would forget him the next day.

Tags

Comments (7)