AI in Rules
complete
James Thurlow-Craig
One use case example:
We ask all customers how their experience was with us. If they respond and it's positive, we send them a link to ask for a review. If it's negative and/or they also have other points raised, we handle those.
If rules had an option to work like below it would be amazing:
OpenAI Query: If contents of email is positive and does not require any further action from us THEN reply with Email Template. If contents of email is negative, reply with Email Template 2 and Label it "To Action".
I'm sure there are many other ways AI in rules could be used, for example to triage support tickets by priority or by best User to respond etc.
Philippe Lehoux
complete
Today, we're launching AI Rules in Missive - a simple way to bring the intelligence of AI to your email workflow without the complexity. It's the same flexible rule engine you love, now with the power to understand email content the way you do.
AI Rules are an extension of Missive's existing rules engine. If you've used rules before, you know they're powerful for automating repetitive tasks based on simple conditions like sender address or subject line keywords.
Now, we're adding the ability to use AI to understand what an email is actually about.
Here's how it works:
1. You create a rule with an "AI Prompt" condition
2. You write a simple instruction for the AI
3. The AI reads the email and gives an answer
4. Your rule takes action based on that answer
No training data. No complicated setup. Just plain language instructions that the AI follows.
Learn more https://missiveapp.com/blog/autopilot-for-your-inbox-ai-rules-have-arrived
Philippe Lehoux
in progress
Arvind Goel
AI-based triage would be a very useful feature.
Philippe Lehoux
Merged in a post:
Use AI to automatically tag emails
Ian Mackey
Use AI to automatically tag emails or suggest new tags
Philippe Lehoux
Merged in a post:
Use AI to help filter emails.
Joey Fenny
I get so many emails every day, I simply cannot keep up. I was thinking about building such a feature myself with a custom server / email monitor, however, after seeing Missive's new OpenAI integration, it dawned on me that maybe it would be easier if Missive had such a feature built in.
Basically, the idea is: Use OpenAI/ChatGPT/GPT-3 or whatever, to scan the email / body of an email, and then decide if that email is worth my time to read or not. If not, send it straight to spam or trash, but don't bother me with it.
Proof of concept:
ChatGPT Prompt:
Respond with nothing other than a boolean based on if the following email is worth my time to read.
It is important to know that emails should be considered not worth my time to read if they are: Not personalized, may appear to be a mass email, is not transactional, is a marketing email, or is overall just spammy.
Worth reading = true
Not worth reading = false
Email subject: Advertising Made Easy on Pinterest
Email body:
Hi,
The holiday season may be over, but many people have received money and gift cards for the holidays!
This is the perfect time to capitalize and target your future customers. They have money to spend and they need a place to spend it.
I am here to help you build your advertising strategy to pull those customers in. Let's get you on a call to discuss the recommendations I have for Electro Gear.
Here is my Calendly to find a time that best works for you.
Response: false
---
The risk with this:
It may false flag some emails as spam and those emails may be important and cause problem if missed.
Resolution:
There will always be a risk to this. But the risk to reward is high. ChatGPT is very smart. It will not often mark emails wrong. To lower the risk, we could not actually delete the emails but move them into the background somewhere
JP Boily
While ChatGPT is not the proper solution due to not being secure (your prompts are part of the ChatGPT history; look at the Samsung board "leak" through ChatGPT), something that would filter and classify emails would be a MASSIVE help.