Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A hackable AI assistant using a single SQLite table and a handful of cron jobs

geoffreylitt.com

Post reply on HN
Print view View original

101–110 of 183 posts

Re: A hackable AI assistant using a single SQLite table and a handful of cron jobs

#101

Post by dogline »

dogline wrote: post #4

This made me think: what if my little utility assistant program that I have, similar to your Stevens, had access to a mailbox? I've got a little utility program that I can tell to get the weather or run common commands unique to my system. It's handy, and I can even cron it to run things regularly, if I'd like. If it had its own email box, I can send it information, it could use AI to parse that info, and possibly se…

*Update*: I tried writing a little Python code to read and write from a mailbox, reading worked great, but writing an email had the email disappear to some filter or spam or something somewhere. I've got to figure out where it went, but this is the warning that some people had about not trusting a messaging protocol (email in this case) when you can't control the servers. Messages can disappear.

I read that [Mailgun](https://www.mailgun.com/) might improve this. Haven't tried it yet.

Other alternatives for messages that I haven't tried. My requirement is to be able to send messages and send/receive on my mobile device. I do not want to write a mobile app.

* [Telegram](https://telegram.org/) (OP's system) with [bots](https://core.telegram.org/bots)

* [MQTT](https://mqtt.org/) with server

* [Notify (ntfy.sh)](https://ntfy.sh/)

* Email (ubiquitous)

   * [Mailgun](https://www.mailgun.com/)

   * [CloudMailin](https://www.cloudmailin.com/)
Also, to [simonw](https://news.ycombinator.com/user?id=simonw) point, LLM calls are cheap now, especially with something as low tokens as this.

And, links don't format in HN markdown. I did the work to include them, they're staying in.

Re: A hackable AI assistant using a single SQLite table and a handful of cron jobs

#102

Post by emporas »

It reminds me of "Generative AI is just a phase. What’s next is interactive AI."

The more i think about it however, command line applications are about as interactive as a program can be.

Let's say one is interested to find out which one of the following 7 next days is gonna rain and send it to a telegram bot. `weather_forecast --days 7 | grep rain` | send_telegram.

The whole thing of off-loading everything to nondeterministic computation instead of the good ol determinism does seem strange to me. I am a huge fan though, of using non-deterministic computation for creating deterministic computation, i.e. programming.

/As a side note, i have played chess against Ishiguro many times on lichess.

[1] https://www.technologyreview.com/2023/09/15/1079624/deepmind...

Re: A hackable AI assistant using a single SQLite table and a handful of cron jobs

#104

Post by jredwards »

I've been kicking around idea for a similar open source project, with the caveats that:

1. I'd like the backend to be configured for any LLM the user might happen to have access to (be that the API for a paid service or something locally hosted on-prem).

2. I'm also wondering how feasible it is to hook it up to a touchscreen running on some hopped-up raspberry pi platform so that it can be interacted with like an Alexa device or any of the similar offerings from other companies. Ideally, that means voice controls as well, which are potentially another technical problem (OpenAI's API will accept an audio file, but for most other services you'd have to do voice to text before sending the prompt off to the API).

3. I'd like to make the integrations extensible. Calendar, weather, but maybe also homebridge, spotify, etc. I'm wondering if MCP servers are the right avenue for that.

I don't have the bandwidth to commit a lot of time to a project like this right now, but if anyone else is charting in this direction I'd love to participate.

Re: A hackable AI assistant using a single SQLite table and a handful of cron jobs

#105

Post by ajcp »

simonw wrote: post #78
ajcp wrote: post #70

I'm a little confused as to the 16-bit game interface shown in the article. Is that just for illustration purposes in the article itself, or is there an actual UI you've built to represent Steven/Steven's world?

It's a real UI - the code for that is here: https://www.val.town/x/geoffreylitt/stevensDemo/code/dashboa...

Thanks for the confirmation! I came across that but am unfamiliar with Val.town so wasn't sure of the repo structure I was looking through.

Re: A hackable AI assistant using a single SQLite table and a handful of cron jobs

#107

Post by panki27 »

jredwards wrote: post #104

I've been kicking around idea for a similar open source project, with the caveats that: 1. I'd like the backend to be configured for any LLM the user might happen to have access to (be that the API for a paid service or something locally hosted on-prem). 2. I'm also wondering how feasible it is to hook it up to a touchscreen running on some hopped-up raspberry pi platform so that it can be interacted with like an Ale…

You might want to take a look at SillyTavern. Supports multiple backends, accepts voice input, and has a plugin system.

Re: A hackable AI assistant using a single SQLite table and a handful of cron jobs

#108

Post by ajcp »

alexchamberlain wrote: post #71
ajcp wrote: post #70

I'm a little confused as to the 16-bit game interface shown in the article. Is that just for illustration purposes in the article itself, or is there an actual UI you've built to represent Steven/Steven's world?

Towards the end of the article, the author implies it is real when they explain why they made it that way (TL;DR: A bit of fun)

I just assumed he was talking about the actual database and LLM setup as the bit of fun :D

Re: A hackable AI assistant using a single SQLite table and a handful of cron jobs

#109

Post by triyambakam »

simonw wrote: post #87
triyambakam wrote: post #84

Earlier quoted context omitted.

No I don't. Do you?

Yes. Trusting them is my competitive advantage. I've also been following Anthropic pretty closely for the last two years and I've seen no evidence that they would break their principles here and plenty of evidence of how far they go to respect the privacy of their users: https://simonwillison.net/2024/Dec/12/clio/

> Yes. Trusting them is my competitive advantage.

I don't see what that is supposed to mean. What does that give you?

Re: A hackable AI assistant using a single SQLite table and a handful of cron jobs

#110

Post by dostick »

panki27 wrote: post #107
jredwards wrote: post #104

I've been kicking around idea for a similar open source project, with the caveats that: 1. I'd like the backend to be configured for any LLM the user might happen to have access to (be that the API for a paid service or something locally hosted on-prem). 2. I'm also wondering how feasible it is to hook it up to a touchscreen running on some hopped-up raspberry pi platform so that it can be interacted with like an Ale…

You might want to take a look at SillyTavern. Supports multiple backends, accepts voice input, and has a plugin system.

I keep hearing about it, but never got to check out, the name suggests that it may be waste of time. Maybe it’s a fantastic project but name lets it down?
Post reply on HN