Interesting choice with telegram. I see it requires to get user's own Gmail api credentials?
I too recently launched AI automations for Gmail. I am a heavy email user, so connecting chatgpt to my inbox was obvious to me. When gpt-4 was released it passed my tests with 90% accuracy, so I've decided to release it at inboxchat.ai
But while I enjoy AI sorting my emails, the most useful feature for meturned out to be screener, inspired by this post: https://blog.nawaz.org/posts/2018/Sep/solving-my-email-probl.... It sends all emails from first-time or unknown senders into a separate Review Later label. If email is moved to Inbox, then that sender becomes trusted and is allowed into Inbox in the future. It's simple, dumb feature but so effective
I would use something like inboxchat.ai, sort of. What I want is to get proposed labels based on my usual labeling activity, and then I want to periodically login and review recommended labels for the activity (unlabeled unread pending email) in my inbox, where I would either speed edit some of the suggestions or accept the recommended labels.
While I have hundreds of auto-label filters already setup, there's always constantly more activity to consider. So having label recommendations could save a lot of time. Or proposed new filters based on common obvious labeling activity already occurring in my inbox.
Telegram seems like strange choice but ok. I personally started to tinker with something similar - IMAP client which learns how you categorize emails into folders and helps with that. I started with manual rules and want to connect some optional automagic later.
I noticed that some emails are valid for only some hours and then it's in inbox unnecessarily giving that +1 to unread badge.
On other hand server side filtering works before arriving in inbox so you don't see for example, that parcel will arrive today. Next day you need to manually archive it. After 10-20 mails like that, if you are busy - half of inbox needlessly needs user action if you are into inbox zero.
ChatGPT's success comes with a chat based UI, every LLM AI agent these days had a hidden chat context somewhere. So choosing a chat app is pretty intuitive and straightforward.
Well that's a valid point, I might add other communication channel later, maybe the community will add a new one ;)
Interesting point - for the moment all actions are manually validated on this project but I might also add features to make the agent autonomous to some extent.
Yeah, I don't trust AI magic either. Very precise manual filters can run out of date quickly, "AI" can be unpredictable. Maybe naïve bayes will be a middle ground?
I use and like SpamSieve for macOS but it's limited to spam/not spam.
I would really like support for more categories than two.
Hi. I use shortwave (AI email built on top of Gmail) but I’m growing frustrated with it. I’ve never really gotten into telegram as an assistant. Would you mind sharing why telegram is the preferred interaction medium of choice here?
On Telegram It's just very fast to create a bot and getting started.
I've tried WhatsApp before but too complicated and Signal was also complex.
I'd say that Telegram was the easiest to get a poc working.
Did you consider Zulip? I've always wondered why Zulip isn't used more often by communities. Seems like it could be a great hybrid between Discord and Telegram.
I too recently launched AI automations for Gmail. I am a heavy email user, so connecting chatgpt to my inbox was obvious to me. When gpt-4 was released it passed my tests with 90% accuracy, so I've decided to release it at inboxchat.ai
But while I enjoy AI sorting my emails, the most useful feature for meturned out to be screener, inspired by this post: https://blog.nawaz.org/posts/2018/Sep/solving-my-email-probl.... It sends all emails from first-time or unknown senders into a separate Review Later label. If email is moved to Inbox, then that sender becomes trusted and is allowed into Inbox in the future. It's simple, dumb feature but so effective