DeepSeek offers an open-source model that is supposed to be able to work offline, local, and is free.
I've never used it, but theoretically it is quite capable. Does it have a running memoy of tasks it will remember for later, or is every session a new blank slate? Does it keep track of the current time and will it wake up and tell you reminders at specific times without you remembering to ask it? Will it be able to guess the current typical weather offline? These are questions I don't know.
Theoretically you can train AI to do what you want it to do. Especially if it is open source and has a community of people testing it out.
Edit:
I asked the online vesion of deepseek some questions. Here is the result:
>do you remember our conversations?
Currently, I don’t have memory retention between conversations—each time you start a new chat, I don’t recall our previous interactions. However, I can reference details you provide in the current conversation to keep things consistent.
If you’d like me to remember something for this session, just let me know, and I’ll do my best to keep track within this chat!
How can I assist you today?
>can you tell me something at a specific time?
Currently, I don’t have the ability to schedule reminders or trigger actions at specific times (like sending a message later). However, you can:
1. **Set your own reminder** (phone alarm, calendar, etc.) and return to chat with me when the time comes.
2. **Ask me something now**, and I’ll respond immediately—for example, a fun fact, a quote, or advice!
What would you like me to share? Or tell me the topic/time, and I’ll help brainstorm solutions!
*(Note: If this is about productivity, I’m happy to suggest tools like Google Calendar, Todoist, or phone alarms!)*
>does your offline model have either of those features?
Great question! My **offline model** (the version you might run locally, like DeepSeek-V3 or similar open-weight models) also **does not** have built-in features for:
- **Scheduled reminders/alerts** (no native ability to "ping" you at a set time).
- **Memory retention** between sessions (unless manually saved/loaded via external tools).
### Why?
1. **No active connectivity**: Offline models lack internet access or background processes to trigger timed actions.
2. **Session-based**: Like the online version, they typically reset memory after each session unless integrated with a custom app/database.
### Workarounds:
- **External tools**: Pair the model with a scheduler (e.g., cron jobs, Tasker, or scripts) to automate prompts at specific times.
- **Manual save/load**: Some local UIs (like Ollama, LM Studio) allow saving chat history to "remember" context across sessions.
Would you like help brainstorming a specific setup? For example, I can suggest how to combine a local model with a calendar API or script!
....
So, the offline model could be tweaked to do it; and odds are someone has already done it and has written a guide.
....
And then I said:
>thank you! i asked because someone was wondering
You're very welcome! I’m glad you asked—it’s a great question, and it helps clarify what’s possible (and what’s still sci-fi… for now!).
If that person (or you!) ever wants to explore creative ways to **approximate** these features—like combining local models with other tools—feel free to ping me. I’d be happy to brainstorm ideas!
Otherwise, hope you both have an awesome day!
*(And remember: I’m always here for instant answers, even if I can’t yet whisper them at 3 AM.)*
....
Ha Ha!
