So I paid for a lifetime version and now realised that if I want to use any basic AI features, I have to pay a monthly fee, which in my opinion is not a good solution, especially as everyone will already pay for AI and it should just be able to implement it. However, I did sign up for months just to find that the AI integration is useless for me. I usually like to jot down some points and then let AI draft a reply, similar to what I do with Canary and Copilot. But I find that EM Client only provides useless outputs for me where it doesn’t understand that it’s replying to an email and instead it replies to me. Maybe I’m not using it correctly, but I don’t find it very helpful.
1 Like
We’ll consider adding an option for third party API keys in the future versions. If eM Client’s AI (based on ChatGPT) does not work well for you, it would be great to post it here or contact support, as we of course test the prompts, but it may not work all the time, so we need your feedback to improve it.
2 Likes
There’s a lot of AI tools emerging and workplaces will be issuing guidance to employees. We’ve been banned from using AI for certain tasks and someone I know has had their work tell them only Microsoft Copilot can be used. We’re likely at a stage where the number of tools will be growing and so employers will be coming up to speed, and I can imagine there will be a clamp down coming after that initial growth, so that employers reduce their risks.
Many AI tools will offer a free service but where they use your queries for training, which means nothing sensitive/confidential should be sent to them. If there’s a chance anything might be sensitive, then it is best to use the paid versions of services, which typically do not train on the user’s data, but that means the user then has their own account with one of these.
Adding a low cost monthly/annual payment (to eM Client) for an option where the user specifies their preferred AI tool would have more potential for me.
For technical areas, I’ve seen people prefer Perplexity over ChatGPT (which has been known to make things up!).
1 Like
I have thought about paying for AI on eM Client but I don’t think it does enough to warrant it.
In Apple Mail, for example, I believe mail is sorted through AI, and it even offers mail summaries so that a large message could be summarized into a single paragraph for quick reading.
Perhaps this feature exists in the Em Client offering, but looking through their summary of what is offered, it’s kind of sparse.
Apple also offers a proofreading and rewriting AI tool system-wide at no charge when composing an email.
Of course, I don’t want to use Apple Mail and I also realize Apple is subsidizing the cost of AI for its users, something eM Client can’t do.
However, for me to pay $2 per month per machine, eM Client has to expand its features with AI compatibility.
Finally, another problem for those of us with multiple machines is that the add-ons to eM Client are getting expensive. If I want fast customer support for bugs that are not my fault, I have to pay about $21 a year for 3 licenses. If I want AI, I believe it’s $2 per license per month for those who have purchased licenses. Nobody wants to purchase a lifetime license only to be faced with subscriptions to supplement customer support and AI capability. But again, I understand AI doesn’t come free these days.
1 Like
I would generally suggest to implement MCP capability (https://modelcontextprotocol.io). That would enable almost any LLM app to reasonably interact with eM Client. I already have viable MCP connector in Perplexity for AppleScipt and the results are freakishly productive.
would be keen to hear how you have implemented this with Apple script and your use case.
Sure…
In short I am in the phase of experimenting with it. What I started with was this: GitHub - supermemoryai/apple-mcp: Collection of apple-native tools for the model context protocol. and Perplexity MacOS app through Node.js.
Now through some more connectors and MCP servers from this list: GitHub - modelcontextprotocol/servers: Model Context Protocol Servers and already submitted site I managed to have LLM powered flows powered by textual prompt (inputted by voice) like:
“Send out emails with meeting minutes from my last call w/(somebody) to call attendees and (my_development_team) and highlight relevant tasks to according team members and attendees for each one.”
Perplexity (in specifically prompted space) reads my Apple Mail inbox, search for relevant email with call transcript/minutes (I use Fathom.video service), reads and uses addresses of attendees, generates minutes for each one of them and send it out, menawhile access my Contacts, search for people tagged accordingly (…my_development_team) composes unique email with highlighted content (based on the context and the Perplexity space prompt) and sed it to them.